jprdonnelly/MachineLearningNotebooks
1
0
Fork 0
mirror of https://github.com/Azure/MachineLearningNotebooks.git synced 2025-12-19 09:07:04 -05:00
Code Issues Projects Releases Wiki Activity

update samples from Release-130 as a part of SDK release

Browse Source
This commit is contained in:
amlrelsa-ms
2022-03-29 22:33:38 +00:00
parent 796798cb49
commit 95b0392ed2
534 changed files with 151904 additions and 27048 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
CODE_OF_CONDUCT.md Normal file
View File

@@ -0,0 +1,9 @@
# Microsoft Open Source Code of Conduct
This project has adopted the [Microsoft Open Source Code of Conduct](https://opensource.microsoft.com/codeofconduct/).
Resources:
- [Microsoft Open Source Code of Conduct](https://opensource.microsoft.com/codeofconduct/)
- [Microsoft Code of Conduct FAQ](https://opensource.microsoft.com/codeofconduct/faq/)
- Contact [opencode@microsoft.com](mailto:opencode@microsoft.com) with questions or concerns

21
LICENSE Normal file
View File

@@ -0,0 +1,21 @@
MIT License
Copyright (c) Microsoft Corporation. All rights reserved.
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE

14
Licenses/sdk-license/LICENSE Normal file
View File

@@ -0,0 +1,14 @@
This software is made available to you on the condition that you agree to
[your agreement][1] governing your use of Azure.
If you do not have an existing agreement governing your use of Azure, you agree that
your agreement governing use of Azure is the [Microsoft Online Subscription Agreement][2]
(which incorporates the [Online Services Terms][3]).
By using the software you agree to these terms. This software may collect data
that is transmitted to Microsoft. Please see the [Microsoft Privacy Statement][4]
to learn more about how Microsoft processes personal data.
[1]: https://azure.microsoft.com/en-us/support/legal/
[2]: https://azure.microsoft.com/en-us/support/legal/subscription-agreement/
[3]: http://www.microsoftvolumelicensing.com/DocumentSearch.aspx?Mode=3&DocumentTypeId=46
[4]: http://go.microsoft.com/fwlink/?LinkId=248681

15
Licenses/sdk-preview-license/LICENSE Normal file
View File

@@ -0,0 +1,15 @@
This Preview is made available to you on the condition that you agree to the
[Supplemental Terms of Use for Microsoft Azure Previews][1], which supplement
[your agreement][2] governing your use of Azure.
If you do not have an existing agreement governing your use of Azure, you agree that
your agreement governing use of Azure is the [Microsoft Online Subscription Agreement][3]
(which incorporates the [Online Services Terms][4]).
By using the Preview you agree to these terms. This Preview may collect data
that is transmitted to Microsoft. Please see the [Microsoft Privacy Statement][5]
to learn more about how Microsoft processes personal data.
[1]: https://azure.microsoft.com/en-us/support/legal/preview-supplemental-terms/
[2]: https://azure.microsoft.com/en-us/support/legal/
[3]: https://azure.microsoft.com/en-us/support/legal/subscription-agreement/
[4]: http://www.microsoftvolumelicensing.com/DocumentSearch.aspx?Mode=3&DocumentTypeId=46
[5]: http://go.microsoft.com/fwlink/?LinkId=248681

95
NBSETUP.md Normal file
View File

@@ -0,0 +1,95 @@
# Set up your notebook environment for Azure Machine Learning
To run the notebooks in this repository use one of following options.
## **Option 1: Use Azure Notebooks**
Azure Notebooks is a hosted Jupyter-based notebook service in the Azure cloud. Azure Machine Learning Python SDK is already pre-installed in the Azure Notebooks `Python 3.6` kernel.
1. [![Azure Notebooks](https://notebooks.azure.com/launch.png)](https://aka.ms/aml-clone-azure-notebooks)
[Import sample notebooks ](https://aka.ms/aml-clone-azure-notebooks) into Azure Notebooks
1. Follow the instructions in the [Configuration](configuration.ipynb) notebook to create and connect to a workspace
1. Open one of the sample notebooks
**Make sure the Azure Notebook kernel is set to `Python 3.6`** when you open a notebook by choosing Kernel > Change Kernel > Python 3.6 from the menus.
## **Option 2: Use your own notebook server**
### Quick installation
We recommend you create a Python virtual environment ([Miniconda](https://conda.io/miniconda.html) preferred but [virtualenv](https://virtualenv.pypa.io/en/latest/) works too) and install the SDK in it.
```sh
# install just the base SDK
pip install azureml-sdk
# clone the sample repoistory
git clone https://github.com/Azure/MachineLearningNotebooks.git
# below steps are optional
# install the base SDK, Jupyter notebook server and tensorboard
pip install azureml-sdk[notebooks,tensorboard]
# install model explainability component
pip install azureml-sdk[interpret]
# install automated ml components
pip install azureml-sdk[automl]
# install experimental features (not ready for production use)
pip install azureml-sdk[contrib]
```
Note the _extras_ (the keywords inside the square brackets) can be combined. For example:
```sh
# install base SDK, Jupyter notebook and automated ml components
pip install azureml-sdk[notebooks,automl]
```
### Full instructions
[Install the Azure Machine Learning SDK](https://docs.microsoft.com/en-us/azure/machine-learning/service/quickstart-create-workspace-with-python)
Please make sure you start with the [Configuration](configuration.ipynb) notebook to create and connect to a workspace.
### Video walkthrough:
[!VIDEO https://youtu.be/VIsXeTuW3FU]
## **Option 3: Use Docker**
You need to have Docker engine installed locally and running. Open a command line window and type the following command.
__Note:__ We use version `1.0.10` below as an exmaple, but you can replace that with any available version number you like.
```sh
# clone the sample repoistory
git clone https://github.com/Azure/MachineLearningNotebooks.git
# change current directory to the folder
# where Dockerfile of the specific SDK version is located.
cd MachineLearningNotebooks/Dockerfiles/1.0.10
# build a Docker image with the a name (azuremlsdk for example)
# and a version number tag (1.0.10 for example).
# this can take several minutes depending on your computer speed and network bandwidth.
docker build . -t azuremlsdk:1.0.10
# launch the built Docker container which also automatically starts
# a Jupyter server instance listening on port 8887 of the host machine
docker run -it -p 8887:8887 azuremlsdk:1.0.10
```
Now you can point your browser to http://localhost:8887. We recommend that you start from the `configuration.ipynb` notebook at the root directory.
If you need additional Azure ML SDK components, you can either modify the Docker files before you build the Docker images to add additional steps, or install them through command line in the live container after you build the Docker image. For example:
```sh
# install the core SDK and automated ml components
pip install azureml-sdk[automl]
# install the core SDK and model explainability component
pip install azureml-sdk[interpret]
# install the core SDK and experimental components
pip install azureml-sdk[contrib]
```
Drag and Drop
The image will be downloaded by Fatkun

43
README.md Normal file
View File

@@ -0,0 +1,43 @@
# Azure Machine Learning Python SDK notebooks
> a community-driven repository of examples using mlflow for tracking can be found at https://github.com/Azure/azureml-examples
Welcome to the Azure Machine Learning Python SDK notebooks repository!
## Getting started
These notebooks are recommended for use in an Azure Machine Learning [Compute Instance](https://docs.microsoft.com/azure/machine-learning/concept-compute-instance), where you can run them without any additional set up.
However, the notebooks can be run in any development environment with the correct `azureml` packages installed.
Install the `azureml.core` Python package:
```sh
pip install azureml-core
```
Install additional packages as needed:
```sh
pip install azureml-mlflow
pip install azureml-dataset-runtime
pip install azureml-automl-runtime
pip install azureml-pipeline
pip install azureml-pipeline-steps
...
```
We recommend starting with one of the [quickstarts](tutorials/compute-instance-quickstarts).
## Contributing
This repository is a push-only mirror. Pull requests are ignored.
## Code of Conduct
This project has adopted the [Microsoft Open Source Code of Conduct](https://opensource.microsoft.com/codeofconduct/). Please see the [code of conduct](CODE_OF_CONDUCT.md) for details.
## Reference
- [Documentation](https://docs.microsoft.com/azure/machine-learning)

772
configuration.ipynb
View File

@@ -1,389 +1,389 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copyright (c) Microsoft Corporation. All rights reserved.\n",
"\n",
"Licensed under the MIT License."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"![Impressions](https://PixelServer20190423114238.azurewebsites.net/api/impressions/MachineLearningNotebooks/configuration.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Configuration\n",
"\n",
"_**Setting up your Azure Machine Learning services workspace and configuring your notebook library**_\n",
"\n",
"---\n",
"---\n",
"\n",
"## Table of Contents\n",
"\n",
"1. [Introduction](#Introduction)\n",
" 1. What is an Azure Machine Learning workspace\n",
"1. [Setup](#Setup)\n",
" 1. Azure subscription\n",
" 1. Azure ML SDK and other library installation\n",
" 1. Azure Container Instance registration\n",
"1. [Configure your Azure ML Workspace](#Configure%20your%20Azure%20ML%20workspace)\n",
" 1. Workspace parameters\n",
" 1. Access your workspace\n",
" 1. Create a new workspace\n",
" 1. Create compute resources\n",
"1. [Next steps](#Next%20steps)\n",
"\n",
"---\n",
"\n",
"## Introduction\n",
"\n",
"This notebook configures your library of notebooks to connect to an Azure Machine Learning (ML) workspace. In this case, a library contains all of the notebooks in the current folder and any nested folders. You can configure this notebook library to use an existing workspace or create a new workspace.\n",
"\n",
"Typically you will need to run this notebook only once per notebook library as all other notebooks will use connection information that is written here. If you want to redirect your notebook library to work with a different workspace, then you should re-run this notebook.\n",
"\n",
"In this notebook you will\n",
"* Learn about getting an Azure subscription\n",
"* Specify your workspace parameters\n",
"* Access or create your workspace\n",
"* Add a default compute cluster for your workspace\n",
"\n",
"### What is an Azure Machine Learning workspace\n",
"\n",
"An Azure ML Workspace is an Azure resource that organizes and coordinates the actions of many other Azure resources to assist in executing and sharing machine learning workflows. In particular, an Azure ML Workspace coordinates storage, databases, and compute resources providing added functionality for machine learning experimentation, deployment, inference, and the monitoring of deployed models."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Setup\n",
"\n",
"This section describes activities required before you can access any Azure ML services functionality."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### 1. Azure Subscription\n",
"\n",
"In order to create an Azure ML Workspace, first you need access to an Azure subscription. An Azure subscription allows you to manage storage, compute, and other assets in the Azure cloud. You can [create a new subscription](https://azure.microsoft.com/en-us/free/) or access existing subscription information from the [Azure portal](https://portal.azure.com). Later in this notebook you will need information such as your subscription ID in order to create and access AML workspaces.\n",
"\n",
"### 2. Azure ML SDK and other library installation\n",
"\n",
"If you are running in your own environment, follow [SDK installation instructions](https://docs.microsoft.com/azure/machine-learning/service/how-to-configure-environment). If you are running in Azure Notebooks or another Microsoft managed environment, the SDK is already installed.\n",
"\n",
"Also install following libraries to your environment. Many of the example notebooks depend on them\n",
"\n",
"```\n",
"(myenv) $ conda install -y matplotlib tqdm scikit-learn\n",
"```\n",
"\n",
"Once installation is complete, the following cell checks the Azure ML SDK version:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"tags": [
"install"
]
},
"outputs": [],
"source": [
"import azureml.core\n",
"\n",
"print(\"This notebook was created using version AZUREML-SDK-VERSION of the Azure ML SDK\")\n",
"print(\"You are currently using version\", azureml.core.VERSION, \"of the Azure ML SDK\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"If you are using an older version of the SDK then this notebook was created using, you should upgrade your SDK.\n",
"\n",
"### 3. Azure Container Instance registration\n",
"Azure Machine Learning uses of [Azure Container Instance (ACI)](https://azure.microsoft.com/services/container-instances) to deploy dev/test web services. An Azure subscription needs to be registered to use ACI. If you or the subscription owner have not yet registered ACI on your subscription, you will need to use the [Azure CLI](https://docs.microsoft.com/en-us/cli/azure/install-azure-cli?view=azure-cli-latest) and execute the following commands. Note that if you ran through the AML [quickstart](https://docs.microsoft.com/en-us/azure/machine-learning/service/quickstart-get-started) you have already registered ACI. \n",
"\n",
"```shell\n",
"# check to see if ACI is already registered\n",
"(myenv) $ az provider show -n Microsoft.ContainerInstance -o table\n",
"\n",
"# if ACI is not registered, run this command.\n",
"# note you need to be the subscription owner in order to execute this command successfully.\n",
"(myenv) $ az provider register -n Microsoft.ContainerInstance\n",
"```\n",
"\n",
"---"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Configure your Azure ML workspace\n",
"\n",
"### Workspace parameters\n",
"\n",
"To use an AML Workspace, you will need to import the Azure ML SDK and supply the following information:\n",
"* Your subscription id\n",
"* A resource group name\n",
"* (optional) The region that will host your workspace\n",
"* A name for your workspace\n",
"\n",
"You can get your subscription ID from the [Azure portal](https://portal.azure.com).\n",
"\n",
"You will also need access to a [_resource group_](https://docs.microsoft.com/en-us/azure/azure-resource-manager/resource-group-overview#resource-groups), which organizes Azure resources and provides a default region for the resources in a group. You can see what resource groups to which you have access, or create a new one in the [Azure portal](https://portal.azure.com). If you don't have a resource group, the create workspace command will create one for you using the name you provide.\n",
"\n",
"The region to host your workspace will be used if you are creating a new workspace. You do not need to specify this if you are using an existing workspace. You can find the list of supported regions [here](https://azure.microsoft.com/en-us/global-infrastructure/services/?products=machine-learning-service). You should pick a region that is close to your location or that contains your data.\n",
"\n",
"The name for your workspace is unique within the subscription and should be descriptive enough to discern among other AML Workspaces. The subscription may be used only by you, or it may be used by your department or your entire enterprise, so choose a name that makes sense for your situation.\n",
"\n",
"The following cell allows you to specify your workspace parameters. This cell uses the python method `os.getenv` to read values from environment variables which is useful for automation. If no environment variable exists, the parameters will be set to the specified default values. \n",
"\n",
"If you ran the Azure Machine Learning [quickstart](https://docs.microsoft.com/en-us/azure/machine-learning/service/quickstart-get-started) in Azure Notebooks, you already have a configured workspace! You can go to your Azure Machine Learning Getting Started library, view *config.json* file, and copy-paste the values for subscription ID, resource group and workspace name below.\n",
"\n",
"Replace the default values in the cell below with your workspace parameters"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"\n",
"subscription_id = os.getenv(\"SUBSCRIPTION_ID\", default=\"<my-subscription-id>\")\n",
"resource_group = os.getenv(\"RESOURCE_GROUP\", default=\"<my-resource-group>\")\n",
"workspace_name = os.getenv(\"WORKSPACE_NAME\", default=\"<my-workspace-name>\")\n",
"workspace_region = os.getenv(\"WORKSPACE_REGION\", default=\"eastus2\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Access your workspace\n",
"\n",
"The following cell uses the Azure ML SDK to attempt to load the workspace specified by your parameters. If this cell succeeds, your notebook library will be configured to access the workspace from all notebooks using the `Workspace.from_config()` method. The cell can fail if the specified workspace doesn't exist or you don't have permissions to access it. "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core import Workspace\n",
"\n",
"try:\n",
" ws = Workspace(subscription_id = subscription_id, resource_group = resource_group, workspace_name = workspace_name)\n",
" # write the details of the workspace to a configuration file to the notebook library\n",
" ws.write_config()\n",
" print(\"Workspace configuration succeeded. Skip the workspace creation steps below\")\n",
"except:\n",
" print(\"Workspace not accessible. Change your parameters or create a new workspace below\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Create a new workspace\n",
"\n",
"If you don't have an existing workspace and are the owner of the subscription or resource group, you can create a new workspace. If you don't have a resource group, the create workspace command will create one for you using the name you provide.\n",
"\n",
"**Note**: As with other Azure services, there are limits on certain resources (for example AmlCompute quota) associated with the Azure ML service. Please read [this article](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-manage-quotas) on the default limits and how to request more quota.\n",
"\n",
"This cell will create an Azure ML workspace for you in a subscription provided you have the correct permissions.\n",
"\n",
"This will fail if:\n",
"* You do not have permission to create a workspace in the resource group\n",
"* You do not have permission to create a resource group if it's non-existing.\n",
"* You are not a subscription owner or contributor and no Azure ML workspaces have ever been created in this subscription\n",
"\n",
"If workspace creation fails, please work with your IT admin to provide you with the appropriate permissions or to provision the required resources.\n",
"\n",
"**Note**: A Basic workspace is created by default. If you would like to create an Enterprise workspace, please specify sku = 'enterprise'.\n",
"Please visit our [pricing page](https://azure.microsoft.com/en-us/pricing/details/machine-learning/) for more details on our Enterprise edition.\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"tags": [
"create workspace"
]
},
"outputs": [],
"source": [
"from azureml.core import Workspace\n",
"\n",
"# Create the workspace using the specified parameters\n",
"ws = Workspace.create(name = workspace_name,\n",
" subscription_id = subscription_id,\n",
" resource_group = resource_group, \n",
" location = workspace_region,\n",
" create_resource_group = True,\n",
" sku = 'basic',\n",
" exist_ok = True)\n",
"ws.get_details()\n",
"\n",
"# write the details of the workspace to a configuration file to the notebook library\n",
"ws.write_config()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Create compute resources for your training experiments\n",
"\n",
"Many of the sample notebooks use Azure ML managed compute (AmlCompute) to train models using a dynamically scalable pool of compute. In this section you will create default compute clusters for use by the other notebooks and any other operations you choose.\n",
"\n",
"> Note that if you have an AzureML Data Scientist role, you will not have permission to create compute resources. Talk to your workspace or IT admin to create the compute targets described in this section, if they do not already exist.\n",
"\n",
"To create a cluster, you need to specify a compute configuration that specifies the type of machine to be used and the scalability behaviors. Then you choose a name for the cluster that is unique within the workspace that can be used to address the cluster later.\n",
"\n",
"The cluster parameters are:\n",
"* vm_size - this describes the virtual machine type and size used in the cluster. All machines in the cluster are the same type. You can get the list of vm sizes available in your region by using the CLI command\n",
"\n",
"```shell\n",
"az vm list-skus -o tsv\n",
"```\n",
"* min_nodes - this sets the minimum size of the cluster. If you set the minimum to 0 the cluster will shut down all nodes while not in use. Setting this number to a value higher than 0 will allow for faster start-up times, but you will also be billed when the cluster is not in use.\n",
"* max_nodes - this sets the maximum size of the cluster. Setting this to a larger number allows for more concurrency and a greater distributed processing of scale-out jobs.\n",
"\n",
"\n",
"To create a **CPU** cluster now, run the cell below. The autoscale settings mean that the cluster will scale down to 0 nodes when inactive and up to 4 nodes when busy."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core.compute import ComputeTarget, AmlCompute\n",
"from azureml.core.compute_target import ComputeTargetException\n",
"\n",
"# Choose a name for your CPU cluster\n",
"cpu_cluster_name = \"cpu-cluster\"\n",
"\n",
"# Verify that cluster does not exist already\n",
"try:\n",
" cpu_cluster = ComputeTarget(workspace=ws, name=cpu_cluster_name)\n",
" print(\"Found existing cpu-cluster\")\n",
"except ComputeTargetException:\n",
" print(\"Creating new cpu-cluster\")\n",
" \n",
" # Specify the configuration for the new cluster\n",
" compute_config = AmlCompute.provisioning_configuration(vm_size=\"STANDARD_D2_V2\",\n",
" min_nodes=0,\n",
" max_nodes=4)\n",
"\n",
" # Create the cluster with the specified name and configuration\n",
" cpu_cluster = ComputeTarget.create(ws, cpu_cluster_name, compute_config)\n",
" \n",
" # Wait for the cluster to complete, show the output log\n",
" cpu_cluster.wait_for_completion(show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"To create a **GPU** cluster, run the cell below. Note that your subscription must have sufficient quota for GPU VMs or the command will fail. To increase quota, see [these instructions](https://docs.microsoft.com/en-us/azure/azure-supportability/resource-manager-core-quotas-request). "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core.compute import ComputeTarget, AmlCompute\n",
"from azureml.core.compute_target import ComputeTargetException\n",
"\n",
"# Choose a name for your GPU cluster\n",
"gpu_cluster_name = \"gpu-cluster\"\n",
"\n",
"# Verify that cluster does not exist already\n",
"try:\n",
" gpu_cluster = ComputeTarget(workspace=ws, name=gpu_cluster_name)\n",
" print(\"Found existing gpu cluster\")\n",
"except ComputeTargetException:\n",
" print(\"Creating new gpu-cluster\")\n",
" \n",
" # Specify the configuration for the new cluster\n",
" compute_config = AmlCompute.provisioning_configuration(vm_size=\"STANDARD_NC6\",\n",
" min_nodes=0,\n",
" max_nodes=4)\n",
" # Create the cluster with the specified name and configuration\n",
" gpu_cluster = ComputeTarget.create(ws, gpu_cluster_name, compute_config)\n",
"\n",
" # Wait for the cluster to complete, show the output log\n",
" gpu_cluster.wait_for_completion(show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"---\n",
"\n",
"## Next steps\n",
"\n",
"In this notebook you configured this notebook library to connect easily to an Azure ML workspace. You can copy this notebook to your own libraries to connect them to you workspace, or use it to bootstrap new workspaces completely.\n",
"\n",
"If you came here from another notebook, you can return there and complete that exercise, or you can try out the [Tutorials](./tutorials) or jump into \"how-to\" notebooks and start creating and deploying models. A good place to start is the [train within notebook](./how-to-use-azureml/training/train-within-notebook) example that walks through a simplified but complete end to end machine learning process."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"authors": [
{
"name": "ninhu"
}
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copyright (c) Microsoft Corporation. All rights reserved.\n",
"\n",
"Licensed under the MIT License."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"![Impressions](https://PixelServer20190423114238.azurewebsites.net/api/impressions/MachineLearningNotebooks/configuration.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Configuration\n",
"\n",
"_**Setting up your Azure Machine Learning services workspace and configuring your notebook library**_\n",
"\n",
"---\n",
"---\n",
"\n",
"## Table of Contents\n",
"\n",
"1. [Introduction](#Introduction)\n",
" 1. What is an Azure Machine Learning workspace\n",
"1. [Setup](#Setup)\n",
" 1. Azure subscription\n",
" 1. Azure ML SDK and other library installation\n",
" 1. Azure Container Instance registration\n",
"1. [Configure your Azure ML Workspace](#Configure%20your%20Azure%20ML%20workspace)\n",
" 1. Workspace parameters\n",
" 1. Access your workspace\n",
" 1. Create a new workspace\n",
" 1. Create compute resources\n",
"1. [Next steps](#Next%20steps)\n",
"\n",
"---\n",
"\n",
"## Introduction\n",
"\n",
"This notebook configures your library of notebooks to connect to an Azure Machine Learning (ML) workspace. In this case, a library contains all of the notebooks in the current folder and any nested folders. You can configure this notebook library to use an existing workspace or create a new workspace.\n",
"\n",
"Typically you will need to run this notebook only once per notebook library as all other notebooks will use connection information that is written here. If you want to redirect your notebook library to work with a different workspace, then you should re-run this notebook.\n",
"\n",
"In this notebook you will\n",
"* Learn about getting an Azure subscription\n",
"* Specify your workspace parameters\n",
"* Access or create your workspace\n",
"* Add a default compute cluster for your workspace\n",
"\n",
"### What is an Azure Machine Learning workspace\n",
"\n",
"An Azure ML Workspace is an Azure resource that organizes and coordinates the actions of many other Azure resources to assist in executing and sharing machine learning workflows. In particular, an Azure ML Workspace coordinates storage, databases, and compute resources providing added functionality for machine learning experimentation, deployment, inference, and the monitoring of deployed models."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Setup\n",
"\n",
"This section describes activities required before you can access any Azure ML services functionality."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### 1. Azure Subscription\n",
"\n",
"In order to create an Azure ML Workspace, first you need access to an Azure subscription. An Azure subscription allows you to manage storage, compute, and other assets in the Azure cloud. You can [create a new subscription](https://azure.microsoft.com/en-us/free/) or access existing subscription information from the [Azure portal](https://portal.azure.com). Later in this notebook you will need information such as your subscription ID in order to create and access AML workspaces.\n",
"\n",
"### 2. Azure ML SDK and other library installation\n",
"\n",
"If you are running in your own environment, follow [SDK installation instructions](https://docs.microsoft.com/azure/machine-learning/service/how-to-configure-environment). If you are running in Azure Notebooks or another Microsoft managed environment, the SDK is already installed.\n",
"\n",
"Also install following libraries to your environment. Many of the example notebooks depend on them\n",
"\n",
"```\n",
"(myenv) $ conda install -y matplotlib tqdm scikit-learn\n",
"```\n",
"\n",
"Once installation is complete, the following cell checks the Azure ML SDK version:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"tags": [
"install"
]
},
"outputs": [],
"source": [
"import azureml.core\n",
"\n",
"print(\"This notebook was created using version 1.40.0 of the Azure ML SDK\")\n",
"print(\"You are currently using version\", azureml.core.VERSION, \"of the Azure ML SDK\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"If you are using an older version of the SDK then this notebook was created using, you should upgrade your SDK.\n",
"\n",
"### 3. Azure Container Instance registration\n",
"Azure Machine Learning uses of [Azure Container Instance (ACI)](https://azure.microsoft.com/services/container-instances) to deploy dev/test web services. An Azure subscription needs to be registered to use ACI. If you or the subscription owner have not yet registered ACI on your subscription, you will need to use the [Azure CLI](https://docs.microsoft.com/en-us/cli/azure/install-azure-cli?view=azure-cli-latest) and execute the following commands. Note that if you ran through the AML [quickstart](https://docs.microsoft.com/en-us/azure/machine-learning/service/quickstart-get-started) you have already registered ACI. \n",
"\n",
"```shell\n",
"# check to see if ACI is already registered\n",
"(myenv) $ az provider show -n Microsoft.ContainerInstance -o table\n",
"\n",
"# if ACI is not registered, run this command.\n",
"# note you need to be the subscription owner in order to execute this command successfully.\n",
"(myenv) $ az provider register -n Microsoft.ContainerInstance\n",
"```\n",
"\n",
"---"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Configure your Azure ML workspace\n",
"\n",
"### Workspace parameters\n",
"\n",
"To use an AML Workspace, you will need to import the Azure ML SDK and supply the following information:\n",
"* Your subscription id\n",
"* A resource group name\n",
"* (optional) The region that will host your workspace\n",
"* A name for your workspace\n",
"\n",
"You can get your subscription ID from the [Azure portal](https://portal.azure.com).\n",
"\n",
"You will also need access to a [_resource group_](https://docs.microsoft.com/en-us/azure/azure-resource-manager/resource-group-overview#resource-groups), which organizes Azure resources and provides a default region for the resources in a group. You can see what resource groups to which you have access, or create a new one in the [Azure portal](https://portal.azure.com). If you don't have a resource group, the create workspace command will create one for you using the name you provide.\n",
"\n",
"The region to host your workspace will be used if you are creating a new workspace. You do not need to specify this if you are using an existing workspace. You can find the list of supported regions [here](https://azure.microsoft.com/en-us/global-infrastructure/services/?products=machine-learning-service). You should pick a region that is close to your location or that contains your data.\n",
"\n",
"The name for your workspace is unique within the subscription and should be descriptive enough to discern among other AML Workspaces. The subscription may be used only by you, or it may be used by your department or your entire enterprise, so choose a name that makes sense for your situation.\n",
"\n",
"The following cell allows you to specify your workspace parameters. This cell uses the python method `os.getenv` to read values from environment variables which is useful for automation. If no environment variable exists, the parameters will be set to the specified default values. \n",
"\n",
"If you ran the Azure Machine Learning [quickstart](https://docs.microsoft.com/en-us/azure/machine-learning/service/quickstart-get-started) in Azure Notebooks, you already have a configured workspace! You can go to your Azure Machine Learning Getting Started library, view *config.json* file, and copy-paste the values for subscription ID, resource group and workspace name below.\n",
"\n",
"Replace the default values in the cell below with your workspace parameters"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"\n",
"subscription_id = os.getenv(\"SUBSCRIPTION_ID\", default=\"<my-subscription-id>\")\n",
"resource_group = os.getenv(\"RESOURCE_GROUP\", default=\"<my-resource-group>\")\n",
"workspace_name = os.getenv(\"WORKSPACE_NAME\", default=\"<my-workspace-name>\")\n",
"workspace_region = os.getenv(\"WORKSPACE_REGION\", default=\"eastus2\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Access your workspace\n",
"\n",
"The following cell uses the Azure ML SDK to attempt to load the workspace specified by your parameters. If this cell succeeds, your notebook library will be configured to access the workspace from all notebooks using the `Workspace.from_config()` method. The cell can fail if the specified workspace doesn't exist or you don't have permissions to access it. "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core import Workspace\n",
"\n",
"try:\n",
" ws = Workspace(subscription_id = subscription_id, resource_group = resource_group, workspace_name = workspace_name)\n",
" # write the details of the workspace to a configuration file to the notebook library\n",
" ws.write_config()\n",
" print(\"Workspace configuration succeeded. Skip the workspace creation steps below\")\n",
"except:\n",
" print(\"Workspace not accessible. Change your parameters or create a new workspace below\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Create a new workspace\n",
"\n",
"If you don't have an existing workspace and are the owner of the subscription or resource group, you can create a new workspace. If you don't have a resource group, the create workspace command will create one for you using the name you provide.\n",
"\n",
"**Note**: As with other Azure services, there are limits on certain resources (for example AmlCompute quota) associated with the Azure ML service. Please read [this article](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-manage-quotas) on the default limits and how to request more quota.\n",
"\n",
"This cell will create an Azure ML workspace for you in a subscription provided you have the correct permissions.\n",
"\n",
"This will fail if:\n",
"* You do not have permission to create a workspace in the resource group\n",
"* You do not have permission to create a resource group if it's non-existing.\n",
"* You are not a subscription owner or contributor and no Azure ML workspaces have ever been created in this subscription\n",
"\n",
"If workspace creation fails, please work with your IT admin to provide you with the appropriate permissions or to provision the required resources.\n",
"\n",
"**Note**: A Basic workspace is created by default. If you would like to create an Enterprise workspace, please specify sku = 'enterprise'.\n",
"Please visit our [pricing page](https://azure.microsoft.com/en-us/pricing/details/machine-learning/) for more details on our Enterprise edition.\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"tags": [
"create workspace"
]
},
"outputs": [],
"source": [
"from azureml.core import Workspace\n",
"\n",
"# Create the workspace using the specified parameters\n",
"ws = Workspace.create(name = workspace_name,\n",
" subscription_id = subscription_id,\n",
" resource_group = resource_group, \n",
" location = workspace_region,\n",
" create_resource_group = True,\n",
" sku = 'basic',\n",
" exist_ok = True)\n",
"ws.get_details()\n",
"\n",
"# write the details of the workspace to a configuration file to the notebook library\n",
"ws.write_config()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Create compute resources for your training experiments\n",
"\n",
"Many of the sample notebooks use Azure ML managed compute (AmlCompute) to train models using a dynamically scalable pool of compute. In this section you will create default compute clusters for use by the other notebooks and any other operations you choose.\n",
"\n",
"> Note that if you have an AzureML Data Scientist role, you will not have permission to create compute resources. Talk to your workspace or IT admin to create the compute targets described in this section, if they do not already exist.\n",
"\n",
"To create a cluster, you need to specify a compute configuration that specifies the type of machine to be used and the scalability behaviors. Then you choose a name for the cluster that is unique within the workspace that can be used to address the cluster later.\n",
"\n",
"The cluster parameters are:\n",
"* vm_size - this describes the virtual machine type and size used in the cluster. All machines in the cluster are the same type. You can get the list of vm sizes available in your region by using the CLI command\n",
"\n",
"```shell\n",
"az vm list-skus -o tsv\n",
"```\n",
"* min_nodes - this sets the minimum size of the cluster. If you set the minimum to 0 the cluster will shut down all nodes while not in use. Setting this number to a value higher than 0 will allow for faster start-up times, but you will also be billed when the cluster is not in use.\n",
"* max_nodes - this sets the maximum size of the cluster. Setting this to a larger number allows for more concurrency and a greater distributed processing of scale-out jobs.\n",
"\n",
"\n",
"To create a **CPU** cluster now, run the cell below. The autoscale settings mean that the cluster will scale down to 0 nodes when inactive and up to 4 nodes when busy."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core.compute import ComputeTarget, AmlCompute\n",
"from azureml.core.compute_target import ComputeTargetException\n",
"\n",
"# Choose a name for your CPU cluster\n",
"cpu_cluster_name = \"cpu-cluster\"\n",
"\n",
"# Verify that cluster does not exist already\n",
"try:\n",
" cpu_cluster = ComputeTarget(workspace=ws, name=cpu_cluster_name)\n",
" print(\"Found existing cpu-cluster\")\n",
"except ComputeTargetException:\n",
" print(\"Creating new cpu-cluster\")\n",
" \n",
" # Specify the configuration for the new cluster\n",
" compute_config = AmlCompute.provisioning_configuration(vm_size=\"STANDARD_D2_V2\",\n",
" min_nodes=0,\n",
" max_nodes=4)\n",
"\n",
" # Create the cluster with the specified name and configuration\n",
" cpu_cluster = ComputeTarget.create(ws, cpu_cluster_name, compute_config)\n",
" \n",
" # Wait for the cluster to complete, show the output log\n",
" cpu_cluster.wait_for_completion(show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"To create a **GPU** cluster, run the cell below. Note that your subscription must have sufficient quota for GPU VMs or the command will fail. To increase quota, see [these instructions](https://docs.microsoft.com/en-us/azure/azure-supportability/resource-manager-core-quotas-request). "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core.compute import ComputeTarget, AmlCompute\n",
"from azureml.core.compute_target import ComputeTargetException\n",
"\n",
"# Choose a name for your GPU cluster\n",
"gpu_cluster_name = \"gpu-cluster\"\n",
"\n",
"# Verify that cluster does not exist already\n",
"try:\n",
" gpu_cluster = ComputeTarget(workspace=ws, name=gpu_cluster_name)\n",
" print(\"Found existing gpu cluster\")\n",
"except ComputeTargetException:\n",
" print(\"Creating new gpu-cluster\")\n",
" \n",
" # Specify the configuration for the new cluster\n",
" compute_config = AmlCompute.provisioning_configuration(vm_size=\"STANDARD_NC6\",\n",
" min_nodes=0,\n",
" max_nodes=4)\n",
" # Create the cluster with the specified name and configuration\n",
" gpu_cluster = ComputeTarget.create(ws, gpu_cluster_name, compute_config)\n",
"\n",
" # Wait for the cluster to complete, show the output log\n",
" gpu_cluster.wait_for_completion(show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"---\n",
"\n",
"## Next steps\n",
"\n",
"In this notebook you configured this notebook library to connect easily to an Azure ML workspace. You can copy this notebook to your own libraries to connect them to you workspace, or use it to bootstrap new workspaces completely.\n",
"\n",
"If you came here from another notebook, you can return there and complete that exercise, or you can try out the [Tutorials](./tutorials) or jump into \"how-to\" notebooks and start creating and deploying models. A good place to start is the [train within notebook](./how-to-use-azureml/training/train-within-notebook) example that walks through a simplified but complete end to end machine learning process."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
"metadata": {
"authors": [
{
"name": "ninhu"
}
],
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.5"
}
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.5"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
"nbformat": 4,
"nbformat_minor": 2
}

305
contrib/RAPIDS/README.md Normal file
View File

@@ -0,0 +1,305 @@
## How to use the RAPIDS on AzureML materials
### Setting up requirements
The material requires the use of the Azure ML SDK and of the Jupyter Notebook Server to run the interactive execution. Please refer to instructions to [setup the environment.](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-configure-environment#local "Local Computer Set Up") Follow the instructions under **Local Computer**, make sure to run the last step: <span style="font-family: Courier New;">pip install \<new package\></span> with <span style="font-family: Courier New;">new package = progressbar2 (pip install progressbar2)</span>
After following the directions, the user should end up setting a conda environment (<span style="font-family: Courier New;">myenv</span>)that can be activated in an Anaconda prompt
The user would also require an Azure Subscription with a Machine Learning Services quota on the desired region for 24 nodes or more (to be able to select a vmSize with 4 GPUs as it is used on the Notebook) on the desired VM family ([NC\_v3](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#ncv3-series), [NC\_v2](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#ncv2-series), [ND](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#nd-series) or [ND_v2](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#ndv2-series-preview)), the specific vmSize to be used within the chosen family would also need to be whitelisted for Machine Learning Services usage.
&nbsp;
### Getting and running the material
Clone the AzureML Notebooks repository in GitHub by running the following command on a local_directory:
* C:\local_directory>git clone https://github.com/Azure/MachineLearningNotebooks.git
On a conda prompt navigate to the local directory, activate the conda environment (<span style="font-family: Courier New;">myenv</span>), where the Azure ML SDK was installed and launch Jupyter Notebook.
* (<span style="font-family: Courier New;">myenv</span>) C:\local_directory>jupyter notebook
From the resulting browser at http://localhost:8888/tree, navigate to the master notebook:
* http://localhost:8888/tree/MachineLearningNotebooks/contrib/RAPIDS/azure-ml-with-nvidia-rapids.ipynb
&nbsp;
The following notebook will appear:
![](imgs/NotebookHome.png)
&nbsp;
### Master Jupyter Notebook
The notebook can be executed interactively step by step, by pressing the Run button (In a red circle in the above image.)
The first couple of functional steps import the necessary AzureML libraries. If you experience any errors please refer back to the [setup the environment.](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-configure-environment#local "Local Computer Set Up") instructions.
&nbsp;
#### Setting up a Workspace
The following step gathers the information necessary to set up a workspace to execute the RAPIDS script. This needs to be done only once, or not at all if you already have a workspace you can use set up on the Azure Portal:
![](imgs/WorkSpaceSetUp.png)
It is important to be sure to set the correct values for the subscription\_id, resource\_group, workspace\_name, and region before executing the step. An example is:
subscription_id = os.environ.get("SUBSCRIPTION_ID", "1358e503-xxxx-4043-xxxx-65b83xxxx32d")
resource_group = os.environ.get("RESOURCE_GROUP", "AML-Rapids-Testing")
workspace_name = os.environ.get("WORKSPACE_NAME", "AML_Rapids_Tester")
workspace_region = os.environ.get("WORKSPACE_REGION", "West US 2")
&nbsp;
The resource\_group and workspace_name could take any value, the region should match the region for which the subscription has the required Machine Learning Services node quota.
The first time the code is executed it will redirect to the Azure Portal to validate subscription credentials. After the workspace is created, its related information is stored on a local file so that this step can be subsequently skipped. The immediate step will just load the saved workspace
![](imgs/saved_workspace.png)
Once a workspace has been created the user could skip its creation and just jump to this step. The configuration file resides in:
* C:\local_directory\\MachineLearningNotebooks\contrib\RAPIDS\aml_config\config.json
&nbsp;
#### Creating an AML Compute Target
Following step, creates an AML Compute Target
![](imgs/target_creation.png)
Parameter vm\_size on function call AmlCompute.provisioning\_configuration() has to be a member of the VM families ([NC\_v3](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#ncv3-series), [NC\_v2](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#ncv2-series), [ND](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#nd-series) or [ND_v2](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#ndv2-series-preview)) that are the ones provided with P40 or V100 GPUs, that are the ones supported by RAPIDS. In this particular case an Standard\_NC24s\_V2 was used.
&nbsp;
If the output of running the step has an error of the form:
![](imgs/targeterror1.png)
It is an indication that even though the subscription has a node quota for VMs for that family, it does not have a node quota for Machine Learning Services for that family.
You will need to request an increase node quota for that family in that region for **Machine Learning Services**.
&nbsp;
Another possible error is the following:
![](imgs/targeterror2.png)
Which indicates that specified vmSize has not been whitelisted for usage on Machine Learning Services and a request to do so should be filled.
The successful creation of the compute target would have an output like the following:
![](imgs/targetsuccess.png)
&nbsp;
#### RAPIDS script uploading and viewing
The next step copies the RAPIDS script process_data.py, which is a slightly modified implementation of the [RAPIDS E2E example](https://github.com/rapidsai/notebooks/blob/master/mortgage/E2E.ipynb), into a script processing folder and it presents its contents to the user. (The script is discussed in the next section in detail).
If the user wants to use a different RAPIDS script, the references to the <span style="font-family: Courier New;">process_data.py</span> script have to be changed
![](imgs/scriptuploading.png)
&nbsp;
#### Data Uploading
The RAPIDS script loads and extracts features from the Fannie Mae’s Mortgage Dataset to train an XGBoost prediction model. The script uses two years of data
The next few steps download and decompress the data and is made available to the script as an [Azure Machine Learning Datastore](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-access-data).
&nbsp;
The following functions are used to download and decompress the input data
![](imgs/dcf1.png)
![](imgs/dcf2.png)
![](imgs/dcf3.png)
![](imgs/dcf4.png)
&nbsp;
The next step uses those functions to download locally file:
http://rapidsai-data.s3-website.us-east-2.amazonaws.com/notebook-mortgage-data/mortgage_2000-2001.tgz'
And to decompress it, into local folder path = .\mortgage_2000-2001
The step takes several minutes, the intermediate outputs provide progress indicators.
![](imgs/downamddecom.png)
&nbsp;
The decompressed data should have the following structure:
* .\mortgage_2000-2001\acq\Acquisition_<year>Q<num>.txt
* .\mortgage_2000-2001\perf\Performance_<year>Q<num>.txt
* .\mortgage_2000-2001\names.csv
The data is divided in partitions that roughly correspond to yearly quarters. RAPIDS includes support for multi-node, multi-GPU deployments, enabling scaling up and out on much larger dataset sizes. The user will be able to verify that the number of partitions that the script is able to process increases with the number of GPUs used. The RAPIDS script is implemented for single-machine scenarios. An example supporting multiple nodes will be published later.
&nbsp;
The next step upload the data into the [Azure Machine Learning Datastore](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-access-data) under reference <span style="font-family: Courier New;">fileroot = mortgage_2000-2001</span>
The step takes several minutes to load the data, the output provides a progress indicator.
![](imgs/datastore.png)
Once the data has been loaded into the Azure Machine LEarning Data Store, in subsequent run, the user can comment out the ds.upload line and just make reference to the <span style="font-family: Courier New;">mortgage_2000-2001</blog> data store reference
&nbsp;
#### Setting up required libraries and environment to run RAPIDS code
There are two options to setup the environment to run RAPIDS code. The following steps shows how to ues a prebuilt conda environment. A recommended alternative is to specify a base Docker image and package dependencies. You can find sample code for that in the notebook.
![](imgs/install2.png)
&nbsp;
#### Wrapper function to submit the RAPIDS script as an Azure Machine Learning experiment
The next step consists of the definition of a wrapper function to be used when the user attempts to run the RAPIDS script with different arguments. It takes as arguments: <span style="font-family: Times New Roman;">*cpu\_training*</span>; a flag that indicates if the run is meant to be processed with CPU-only, <span style="font-family: Times New Roman;">*gpu\_count*</span>; the number of GPUs to be used if they are meant to be used and part_count: the number of data partitions to be used
![](imgs/wrapper.png)
&nbsp;
The core of the function resides in configuring the run by the instantiation of a ScriptRunConfig object, which defines the source_directory for the script to be executed, the name of the script and the arguments to be passed to the script.
In addition to the wrapper function arguments, two other arguments are passed: <span style="font-family: Times New Roman;">*data\_dir*</span>, the directory where the data is stored and <span style="font-family: Times New Roman;">*end_year*</span> is the largest year to use partition from.
As mentioned earlier the size of the data that can be processed increases with the number of gpus, in the function, dictionary <span style="font-family: Times New Roman;">*max\_gpu\_count\_data\_partition_mapping*</span> maps the maximum number of partitions that we empirically found that the system can handle given the number of GPUs used. The function throws a warning when the number of partitions for a given number of gpus exceeds the maximum but the script is still executed, however the user should expect an error as an out of memory situation would be encountered
If the user wants to use a different RAPIDS script, the reference to the process_data.py script has to be changed
&nbsp;
#### Submitting Experiments
We are ready to submit experiments: launching the RAPIDS script with different sets of parameters.
&nbsp;
The following couple of steps submit experiments under different conditions.
![](imgs/submission1.png)
&nbsp;
The user can change variable num\_gpu between one and the number of GPUs supported by the chosen vmSize. Variable part\_count can take any value between 1 and 11, but if it exceeds the maximum for num_gpu, the run would result in an error
&nbsp;
If the experiment is successfully submitted, it would be placed on a queue for processing, its status would appeared as Queued and an output like the following would appear
![](imgs/queue.png)
&nbsp;
When the experiment starts running, its status would appeared as Running and the output would change to something like this:
![](imgs/running.png)
&nbsp;
#### Reproducing the performance gains plot results on the Blog Post
When the run has finished successfully, its status would appeared as Completed and the output would change to something like this:
&nbsp;
![](imgs/completed.png)
Which is the output for an experiment run with three partitions and one GPU, notice that the reported processing time is 49.16 seconds just as depicted on the performance gains plot on the blog post
&nbsp;
![](imgs/2GPUs.png)
This output corresponds to a run with three partitions and two GPUs, notice that the reported processing time is 37.50 seconds just as depicted on the performance gains plot on the blog post
&nbsp;
![](imgs/3GPUs.png)
This output corresponds to an experiment run with three partitions and three GPUs, notice that the reported processing time is 24.40 seconds just as depicted on the performance gains plot on the blog post
&nbsp;
![](imgs/4gpus.png)
This output corresponds to an experiment run with three partitions and four GPUs, notice that the reported processing time is 23.33 seconds just as depicted on the performance gains plot on the blogpost
&nbsp;
![](imgs/CPUBase.png)
This output corresponds to an experiment run with three partitions and using only CPU, notice that the reported processing time is 9 minutes and 1.21 seconds or 541.21 second just as depicted on the performance gains plot on the blog post
&nbsp;
![](imgs/OOM.png)
This output corresponds to an experiment run with nine partitions and four GPUs, notice that the notebook throws a warning signaling that the number of partitions exceed the maximum that the system can handle with those many GPUs and the run ends up failing, hence having and status of Failed.
&nbsp;
##### Freeing Resources
In the last step the notebook deletes the compute target. (This step is optional especially if the min_nodes in the cluster is set to 0 with which the cluster will scale down to 0 nodes when there is no usage.)
![](imgs/clusterdelete.png)
&nbsp;
### RAPIDS Script
The Master Notebook runs experiments by launching a RAPIDS script with different sets of parameters. In this section, the RAPIDS script, process_data.py in the material, is analyzed
The script first imports all the necessary libraries and parses the arguments passed by the Master Notebook.
The all internal functions to be used by the script are defined.
&nbsp;
#### Wrapper Auxiliary Functions:
The below functions are wrappers for a configuration module for librmm, the RAPIDS Memory Manager python interface:
![](imgs/wap1.png)![](imgs/wap2.png)
&nbsp;
A couple of other functions are wrappers for the submission of jobs to the DASK client:
![](imgs/wap3.png)
![](imgs/wap4.png)
&nbsp;
#### Data Loading Functions:
The data is loaded through the use of the following three functions
![](imgs/DLF1.png)![](imgs/DLF2.png)![](imgs/DLF3.png)
All three functions use library function cudf.read_csv(), cuDF version for the well known counterpart on Pandas.
&nbsp;
#### Data Transformation and Feature Extraction Functions:
The raw data is transformed and processed to extract features by joining, slicing, grouping, aggregating, factoring, etc, the original dataframes just as is done with Pandas. The following functions in the script are used for that purpose:
![](imgs/fef1.png)![](imgs/fef2.png)![](imgs/fef3.png)![](imgs/fef4.png)![](imgs/fef5.png)
![](imgs/fef6.png)![](imgs/fef7.png)![](imgs/fef8.png)![](imgs/fef9.png)
&nbsp;
#### Main() Function
The previous functions are used in the Main function to accomplish several steps: Set up the Dask client, do all ETL operations, set up and train an XGBoost model, the function also assigns which data needs to be processed by each Dask client
&nbsp;
##### Setting Up DASK client:
The following lines:
![](imgs/daskini.png)
&nbsp;
Initialize and set up a DASK client with a number of workers corresponding to the number of GPUs to be used on the run. A successful execution of the set up will result on the following output:
![](imgs/daskoutput.png)
##### All ETL functions are used on single calls to process\_quarter_gpu, one per data partition
![](imgs/ETL.png)
&nbsp;
##### Concentrating the data assigned to each DASK worker
The partitions assigned to each worker are concatenated and set up for training.
![](imgs/Dask2.png)
&nbsp;
##### Setting Training Parameters
The parameters used for the training of a gradient boosted decision tree model are set up in the following code block:
![](imgs/PArameters.png)
Notice how the parameters are modified when using the CPU-only mode.
&nbsp;
##### Launching the training of a gradient boosted decision tree model using XGBoost.
![](imgs/training.png)
The outputs of the script can be observed in the master notebook as the script is executed

547
contrib/RAPIDS/azure-ml-with-nvidia-rapids.ipynb Normal file
View File

@@ -0,0 +1,547 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copyright (c) Microsoft Corporation. All rights reserved.\n",
"\n",
"Licensed under the MIT License."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"![Impressions](https://PixelServer20190423114238.azurewebsites.net/api/impressions/MachineLearningNotebooks/contrib/RAPIDS/azure-ml-with-nvidia-rapids/azure-ml-with-nvidia-rapids.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# NVIDIA RAPIDS in Azure Machine Learning"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The [RAPIDS](https://www.developer.nvidia.com/rapids) suite of software libraries from NVIDIA enables the execution of end-to-end data science and analytics pipelines entirely on GPUs. In many machine learning projects, a significant portion of the model training time is spent in setting up the data; this stage of the process is known as Extraction, Transformation and Loading, or ETL. By using the DataFrame API for ETL\u00c2\u00a0and GPU-capable ML algorithms in RAPIDS, data preparation and training models can be done in GPU-accelerated end-to-end pipelines without incurring serialization costs between the pipeline stages. This notebook demonstrates how to use NVIDIA RAPIDS to prepare data and train model\u00c3\u201a\u00c2\u00a0in Azure.\n",
" \n",
"In this notebook, we will do the following:\n",
" \n",
"* Create an Azure Machine Learning Workspace\n",
"* Create an AMLCompute target\n",
"* Use a script to process our data and train a model\n",
"* Obtain the data required to run this sample\n",
"* Create an AML run configuration to launch a machine learning job\n",
"* Run the script to prepare data for training and train the model\n",
" \n",
"Prerequisites:\n",
"* An Azure subscription to create a Machine Learning Workspace\n",
"* Familiarity with the Azure ML SDK (refer to [notebook samples](https://github.com/Azure/MachineLearningNotebooks))\n",
"* A Jupyter notebook environment with Azure Machine Learning SDK installed. Refer to instructions to [setup the environment](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-configure-environment#local)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Verify if Azure ML SDK is installed"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import azureml.core\n",
"print(\"SDK version:\", azureml.core.VERSION)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"from azureml.core import Workspace, Experiment\n",
"from azureml.core.conda_dependencies import CondaDependencies\n",
"from azureml.core.compute import AmlCompute, ComputeTarget\n",
"from azureml.data.data_reference import DataReference\n",
"from azureml.core.runconfig import RunConfiguration\n",
"from azureml.core import ScriptRunConfig\n",
"from azureml.widgets import RunDetails"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Create Azure ML Workspace"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The following step is optional if you already have a workspace. If you want to use an existing workspace, then\n",
"skip this workspace creation step and move on to the next step to load the workspace.\n",
" \n",
"<font color='red'>Important</font>: in the code cell below, be sure to set the correct values for the subscription_id, \n",
"resource_group, workspace_name, region before executing this code cell."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"subscription_id = os.environ.get(\"SUBSCRIPTION_ID\", \"<subscription_id>\")\n",
"resource_group = os.environ.get(\"RESOURCE_GROUP\", \"<resource_group>\")\n",
"workspace_name = os.environ.get(\"WORKSPACE_NAME\", \"<workspace_name>\")\n",
"workspace_region = os.environ.get(\"WORKSPACE_REGION\", \"<region>\")\n",
"\n",
"ws = Workspace.create(workspace_name, subscription_id=subscription_id, resource_group=resource_group, location=workspace_region)\n",
"\n",
"# write config to a local directory for future use\n",
"ws.write_config()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Load existing Workspace"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"ws = Workspace.from_config()\n",
"\n",
"# if a locally-saved configuration file for the workspace is not available, use the following to load workspace\n",
"# ws = Workspace(subscription_id=subscription_id, resource_group=resource_group, workspace_name=workspace_name)\n",
"\n",
"print('Workspace name: ' + ws.name, \n",
" 'Azure region: ' + ws.location, \n",
" 'Subscription id: ' + ws.subscription_id, \n",
" 'Resource group: ' + ws.resource_group, sep = '\\n')\n",
"\n",
"scripts_folder = \"scripts_folder\"\n",
"\n",
"if not os.path.isdir(scripts_folder):\n",
" os.mkdir(scripts_folder)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Create AML Compute Target"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Because NVIDIA RAPIDS requires P40 or V100 GPUs, the user needs to specify compute targets from one of [NC_v3](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#ncv3-series), [NC_v2](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#ncv2-series), [ND](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#nd-series) or [ND_v2](https://docs.microsoft.com/en-us/azure/virtual-machines/windows/sizes-gpu#ndv2-series-preview) virtual machine types in Azure; these are the families of virtual machines in Azure that are provisioned with these GPUs.\n",
" \n",
"Pick one of the supported VM SKUs based on the number of GPUs you want to use for ETL and training in RAPIDS.\n",
" \n",
"The script in this notebook is implemented for single-machine scenarios. An example supporting multiple nodes will be published later."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"gpu_cluster_name = \"gpucluster\"\n",
"\n",
"if gpu_cluster_name in ws.compute_targets:\n",
" gpu_cluster = ws.compute_targets[gpu_cluster_name]\n",
" if gpu_cluster and type(gpu_cluster) is AmlCompute:\n",
" print('Found compute target. Will use {0} '.format(gpu_cluster_name))\n",
"else:\n",
" print(\"creating new cluster\")\n",
" # vm_size parameter below could be modified to one of the RAPIDS-supported VM types\n",
" provisioning_config = AmlCompute.provisioning_configuration(vm_size = \"Standard_NC6s_v2\", min_nodes=1, max_nodes = 1)\n",
"\n",
" # create the cluster\n",
" gpu_cluster = ComputeTarget.create(ws, gpu_cluster_name, provisioning_config)\n",
" gpu_cluster.wait_for_completion(show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Script to process data and train model"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# copy process_data.py into the script folder\n",
"import shutil\n",
"shutil.copy('./process_data.py', os.path.join(scripts_folder, 'process_data.py'))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Data required to run this sample"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This sample uses [Fannie Mae's Single-Family Loan Performance Data](http://www.fanniemae.com/portal/funding-the-market/data/loan-performance-data.html). Once you obtain access to the data, you will need to make this data available in an [Azure Machine Learning Datastore](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-access-data), for use in this sample. The following code shows how to do that."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Downloading Data"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import tarfile\n",
"import hashlib\n",
"from urllib.request import urlretrieve\n",
"\n",
"def validate_downloaded_data(path):\n",
" if(os.path.isdir(path) and os.path.exists(path + '//names.csv')) :\n",
" if(os.path.isdir(path + '//acq' ) and len(os.listdir(path + '//acq')) == 8):\n",
" if(os.path.isdir(path + '//perf' ) and len(os.listdir(path + '//perf')) == 11):\n",
" print(\"Data has been downloaded and decompressed at: {0}\".format(path))\n",
" return True\n",
" print(\"Data has not been downloaded and decompressed\")\n",
" return False\n",
"\n",
"def show_progress(count, block_size, total_size):\n",
" global pbar\n",
" global processed\n",
" \n",
" if count == 0:\n",
" pbar = ProgressBar(maxval=total_size)\n",
" processed = 0\n",
" \n",
" processed += block_size\n",
" processed = min(processed,total_size)\n",
" pbar.update(processed)\n",
"\n",
" \n",
"def download_file(fileroot):\n",
" filename = fileroot + '.tgz'\n",
" if(not os.path.exists(filename) or hashlib.md5(open(filename, 'rb').read()).hexdigest() != '82dd47135053303e9526c2d5c43befd5' ):\n",
" url_format = 'http://rapidsai-data.s3-website.us-east-2.amazonaws.com/notebook-mortgage-data/{0}.tgz'\n",
" url = url_format.format(fileroot)\n",
" print(\"...Downloading file :{0}\".format(filename))\n",
" urlretrieve(url, filename)\n",
" pbar.finish()\n",
" print(\"...File :{0} finished downloading\".format(filename))\n",
" else:\n",
" print(\"...File :{0} has been downloaded already\".format(filename))\n",
" return filename\n",
"\n",
"def decompress_file(filename,path):\n",
" tar = tarfile.open(filename)\n",
" print(\"...Getting information from {0} about files to decompress\".format(filename))\n",
" members = tar.getmembers()\n",
" numFiles = len(members)\n",
" so_far = 0\n",
" for member_info in members:\n",
" tar.extract(member_info,path=path)\n",
" so_far += 1\n",
" print(\"...All {0} files have been decompressed\".format(numFiles))\n",
" tar.close()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"fileroot = 'mortgage_2000-2001'\n",
"path = '.\\\\{0}'.format(fileroot)\n",
"pbar = None\n",
"processed = 0\n",
"\n",
"if(not validate_downloaded_data(path)):\n",
" print(\"Downloading and Decompressing Input Data\")\n",
" filename = download_file(fileroot)\n",
" decompress_file(filename,path)\n",
" print(\"Input Data has been Downloaded and Decompressed\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Uploading Data to Workspace"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"ds = ws.get_default_datastore()\n",
"\n",
"# download and uncompress data in a local directory before uploading to data store\n",
"# directory specified in src_dir parameter below should have the acq, perf directories with data and names.csv file\n",
"\n",
"# ---->>>> UNCOMMENT THE BELOW LINE TO UPLOAD YOUR DATA IF NOT DONE SO ALREADY <<<<----\n",
"# ds.upload(src_dir=path, target_path=fileroot, overwrite=True, show_progress=True)\n",
"\n",
"# data already uploaded to the datastore\n",
"data_ref = DataReference(data_reference_name='data', datastore=ds, path_on_datastore=fileroot)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Create AML run configuration to launch a machine learning job"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"RunConfiguration is used to submit jobs to Azure Machine Learning service. When creating RunConfiguration for a job, users can either \n",
"1. specify a Docker image with prebuilt conda environment and use it without any modifications to run the job, or \n",
"2. specify a Docker image as the base image and conda or pip packages as dependnecies to let AML build a new Docker image with a conda environment containing specified dependencies to use in the job\n",
"\n",
"The second option is the recommended option in AML. \n",
"The following steps have code for both options. You can pick the one that is more appropriate for your requirements. "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Specify prebuilt conda environment"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The following code shows how to install RAPIDS using conda. The `rapids.yml` file contains the list of packages necessary to run this tutorial. **NOTE:** Initial build of the image might take up to 20 minutes as the service needs to build and cache the new image; once the image is built the subequent runs use the cached image and the overhead is minimal."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"cd = CondaDependencies(conda_dependencies_file_path='rapids.yml')\n",
"run_config = RunConfiguration(conda_dependencies=cd)\n",
"run_config.framework = 'python'\n",
"run_config.target = gpu_cluster_name\n",
"run_config.environment.docker.enabled = True\n",
"run_config.environment.docker.gpu_support = True\n",
"run_config.environment.docker.base_image = \"mcr.microsoft.com/azureml/openmpi4.1.0-cuda11.1-cudnn8-ubuntu20.04\"\n",
"run_config.environment.spark.precache_packages = False\n",
"run_config.data_references={'data':data_ref.to_config()}"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Using Docker"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Alternatively, you can specify RAPIDS Docker image."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# run_config = RunConfiguration()\n",
"# run_config.framework = 'python'\n",
"# run_config.environment.python.user_managed_dependencies = True\n",
"# run_config.environment.python.interpreter_path = '/conda/envs/rapids/bin/python'\n",
"# run_config.target = gpu_cluster_name\n",
"# run_config.environment.docker.enabled = True\n",
"# run_config.environment.docker.gpu_support = True\n",
"# run_config.environment.docker.base_image = \"rapidsai/rapidsai:cuda9.2-runtime-ubuntu18.04\"\n",
"# # run_config.environment.docker.base_image_registry.address = '<registry_url>' # not required if the base_image is in Docker hub\n",
"# # run_config.environment.docker.base_image_registry.username = '<user_name>' # needed only for private images\n",
"# # run_config.environment.docker.base_image_registry.password = '<password>' # needed only for private images\n",
"# run_config.environment.spark.precache_packages = False\n",
"# run_config.data_references={'data':data_ref.to_config()}"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Wrapper function to submit Azure Machine Learning experiment"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# parameter cpu_predictor indicates if training should be done on CPU. If set to true, GPUs are used *only* for ETL and *not* for training\n",
"# parameter num_gpu indicates number of GPUs to use among the GPUs available in the VM for ETL and if cpu_predictor is false, for training as well \n",
"def run_rapids_experiment(cpu_training, gpu_count, part_count):\n",
" # any value between 1-4 is allowed here depending the type of VMs available in gpu_cluster\n",
" if gpu_count not in [1, 2, 3, 4]:\n",
" raise Exception('Value specified for the number of GPUs to use {0} is invalid'.format(gpu_count))\n",
"\n",
" # following data partition mapping is empirical (specific to GPUs used and current data partitioning scheme) and may need to be tweaked\n",
" max_gpu_count_data_partition_mapping = {1: 3, 2: 4, 3: 6, 4: 8}\n",
" \n",
" if part_count > max_gpu_count_data_partition_mapping[gpu_count]:\n",
" print(\"Too many partitions for the number of GPUs, exceeding memory threshold\")\n",
" \n",
" if part_count > 11:\n",
" print(\"Warning: Maximum number of partitions available is 11\")\n",
" part_count = 11\n",
" \n",
" end_year = 2000\n",
" \n",
" if part_count > 4:\n",
" end_year = 2001 # use more data with more GPUs\n",
"\n",
" src = ScriptRunConfig(source_directory=scripts_folder, \n",
" script='process_data.py', \n",
" arguments = ['--num_gpu', gpu_count, '--data_dir', str(data_ref),\n",
" '--part_count', part_count, '--end_year', end_year,\n",
" '--cpu_predictor', cpu_training\n",
" ],\n",
" run_config=run_config\n",
" )\n",
"\n",
" exp = Experiment(ws, 'rapidstest')\n",
" run = exp.submit(config=src)\n",
" RunDetails(run).show()\n",
" return run"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Submit experiment (ETL & training on GPU)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"cpu_predictor = False\n",
"# the value for num_gpu should be less than or equal to the number of GPUs available in the VM\n",
"num_gpu = 1\n",
"data_part_count = 1\n",
"# train using CPU, use GPU for both ETL and training\n",
"run = run_rapids_experiment(cpu_predictor, num_gpu, data_part_count)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Submit experiment (ETL on GPU, training on CPU)\n",
"\n",
"To observe performance difference between GPU-accelerated RAPIDS based training with CPU-only training, set 'cpu_predictor' predictor to 'True' and rerun the experiment"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"cpu_predictor = True\n",
"# the value for num_gpu should be less than or equal to the number of GPUs available in the VM\n",
"num_gpu = 1\n",
"data_part_count = 1\n",
"# train using CPU, use GPU for ETL\n",
"run = run_rapids_experiment(cpu_predictor, num_gpu, data_part_count)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Delete cluster"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# delete the cluster\n",
"# gpu_cluster.delete()"
]
}
],
"metadata": {
"authors": [
{
"name": "ksivas"
}
],
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.8"
}
},
"nbformat": 4,
"nbformat_minor": 4
}

BIN
contrib/RAPIDS/imgs/2GPUs.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 180 KiB

BIN
contrib/RAPIDS/imgs/3GPUs.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 183 KiB

BIN
contrib/RAPIDS/imgs/4gpus.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 183 KiB

BIN
contrib/RAPIDS/imgs/CPUBase.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 177 KiB

BIN
contrib/RAPIDS/imgs/DLF1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 5.0 KiB

BIN
contrib/RAPIDS/imgs/DLF2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 4.8 KiB

BIN
contrib/RAPIDS/imgs/DLF3.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 3.2 KiB

BIN
contrib/RAPIDS/imgs/Dask2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 70 KiB

BIN
contrib/RAPIDS/imgs/ETL.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 64 KiB

BIN
contrib/RAPIDS/imgs/NotebookHome.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 554 KiB

BIN
contrib/RAPIDS/imgs/OOM.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 213 KiB

BIN
contrib/RAPIDS/imgs/PArameters.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 58 KiB

BIN
contrib/RAPIDS/imgs/WorkSpaceSetUp.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 34 KiB

BIN
contrib/RAPIDS/imgs/clusterdelete.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 4.5 KiB

BIN
contrib/RAPIDS/imgs/completed.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 187 KiB

BIN
contrib/RAPIDS/imgs/daskini.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 22 KiB

BIN
contrib/RAPIDS/imgs/daskoutput.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 9.7 KiB

BIN
contrib/RAPIDS/imgs/datastore.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 163 KiB

BIN
contrib/RAPIDS/imgs/dcf1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 3.5 KiB

BIN
contrib/RAPIDS/imgs/dcf2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 2.9 KiB

BIN
contrib/RAPIDS/imgs/dcf3.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 2.5 KiB

BIN
contrib/RAPIDS/imgs/dcf4.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 3.0 KiB

BIN
contrib/RAPIDS/imgs/downamddecom.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 60 KiB

BIN
contrib/RAPIDS/imgs/fef1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 3.5 KiB

BIN
contrib/RAPIDS/imgs/fef2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 3.9 KiB

BIN
contrib/RAPIDS/imgs/fef3.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 5.0 KiB

BIN
contrib/RAPIDS/imgs/fef4.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 4.0 KiB

BIN
contrib/RAPIDS/imgs/fef5.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 4.1 KiB

BIN
contrib/RAPIDS/imgs/fef6.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 4.5 KiB

BIN
contrib/RAPIDS/imgs/fef7.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 5.1 KiB

BIN
contrib/RAPIDS/imgs/fef8.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 3.9 KiB

BIN
contrib/RAPIDS/imgs/fef9.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 3.6 KiB

BIN
contrib/RAPIDS/imgs/install2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 120 KiB

BIN
contrib/RAPIDS/imgs/installation.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 55 KiB

BIN
contrib/RAPIDS/imgs/queue.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 52 KiB

BIN
contrib/RAPIDS/imgs/running.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 181 KiB

BIN
contrib/RAPIDS/imgs/saved_workspace.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 36 KiB

BIN
contrib/RAPIDS/imgs/scriptuploading.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 21 KiB

BIN
contrib/RAPIDS/imgs/submission1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 19 KiB

BIN
contrib/RAPIDS/imgs/target_creation.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 45 KiB

BIN
contrib/RAPIDS/imgs/targeterror1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 31 KiB

BIN
contrib/RAPIDS/imgs/targeterror2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 29 KiB

BIN
contrib/RAPIDS/imgs/targetsuccess.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 10 KiB

BIN
contrib/RAPIDS/imgs/training.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 18 KiB

BIN
contrib/RAPIDS/imgs/wap1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 2.4 KiB

BIN
contrib/RAPIDS/imgs/wap2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 2.5 KiB

BIN
contrib/RAPIDS/imgs/wap3.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 3.4 KiB

BIN
contrib/RAPIDS/imgs/wap4.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 4.8 KiB

BIN
contrib/RAPIDS/imgs/wrapper.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 99 KiB

470
contrib/RAPIDS/process_data.py Normal file
View File

@@ -0,0 +1,470 @@
import numpy as np
import datetime
import dask_xgboost as dxgb_gpu
import dask
import dask_cudf
from dask_cuda import LocalCUDACluster
from dask.delayed import delayed
from dask.distributed import Client, wait
import xgboost as xgb
import cudf
from cudf.dataframe import DataFrame
from collections import OrderedDict
import gc
from glob import glob
import os
import argparse
def run_dask_task(func, **kwargs):
task = func(**kwargs)
return task
def process_quarter_gpu(client, col_names_path, acq_data_path, year=2000, quarter=1, perf_file=""):
dask_client = client
ml_arrays = run_dask_task(delayed(run_gpu_workflow),
col_path=col_names_path,
acq_path=acq_data_path,
quarter=quarter,
year=year,
perf_file=perf_file)
return dask_client.compute(ml_arrays,
optimize_graph=False,
fifo_timeout="0ms")
def null_workaround(df, **kwargs):
for column, data_type in df.dtypes.items():
if str(data_type) == "category":
df[column] = df[column].astype('int32').fillna(-1)
if str(data_type) in ['int8', 'int16', 'int32', 'int64', 'float32', 'float64']:
df[column] = df[column].fillna(-1)
return df
def run_gpu_workflow(col_path, acq_path, quarter=1, year=2000, perf_file="", **kwargs):
names = gpu_load_names(col_path=col_path)
acq_gdf = gpu_load_acquisition_csv(acquisition_path= acq_path + "/Acquisition_"
+ str(year) + "Q" + str(quarter) + ".txt")
acq_gdf = acq_gdf.merge(names, how='left', on=['seller_name'])
acq_gdf.drop_column('seller_name')
acq_gdf['seller_name'] = acq_gdf['new']
acq_gdf.drop_column('new')
perf_df_tmp = gpu_load_performance_csv(perf_file)
gdf = perf_df_tmp
everdf = create_ever_features(gdf)
delinq_merge = create_delinq_features(gdf)
everdf = join_ever_delinq_features(everdf, delinq_merge)
del(delinq_merge)
joined_df = create_joined_df(gdf, everdf)
testdf = create_12_mon_features(joined_df)
joined_df = combine_joined_12_mon(joined_df, testdf)
del(testdf)
perf_df = final_performance_delinquency(gdf, joined_df)
del(gdf, joined_df)
final_gdf = join_perf_acq_gdfs(perf_df, acq_gdf)
del(perf_df)
del(acq_gdf)
final_gdf = last_mile_cleaning(final_gdf)
return final_gdf
def gpu_load_performance_csv(performance_path, **kwargs):
""" Loads performance data
Returns
-------
GPU DataFrame
"""
cols = [
"loan_id", "monthly_reporting_period", "servicer", "interest_rate", "current_actual_upb",
"loan_age", "remaining_months_to_legal_maturity", "adj_remaining_months_to_maturity",
"maturity_date", "msa", "current_loan_delinquency_status", "mod_flag", "zero_balance_code",
"zero_balance_effective_date", "last_paid_installment_date", "foreclosed_after",
"disposition_date", "foreclosure_costs", "prop_preservation_and_repair_costs",
"asset_recovery_costs", "misc_holding_expenses", "holding_taxes", "net_sale_proceeds",
"credit_enhancement_proceeds", "repurchase_make_whole_proceeds", "other_foreclosure_proceeds",
"non_interest_bearing_upb", "principal_forgiveness_upb", "repurchase_make_whole_proceeds_flag",
"foreclosure_principal_write_off_amount", "servicing_activity_indicator"
]
dtypes = OrderedDict([
("loan_id", "int64"),
("monthly_reporting_period", "date"),
("servicer", "category"),
("interest_rate", "float64"),
("current_actual_upb", "float64"),
("loan_age", "float64"),
("remaining_months_to_legal_maturity", "float64"),
("adj_remaining_months_to_maturity", "float64"),
("maturity_date", "date"),
("msa", "float64"),
("current_loan_delinquency_status", "int32"),
("mod_flag", "category"),
("zero_balance_code", "category"),
("zero_balance_effective_date", "date"),
("last_paid_installment_date", "date"),
("foreclosed_after", "date"),
("disposition_date", "date"),
("foreclosure_costs", "float64"),
("prop_preservation_and_repair_costs", "float64"),
("asset_recovery_costs", "float64"),
("misc_holding_expenses", "float64"),
("holding_taxes", "float64"),
("net_sale_proceeds", "float64"),
("credit_enhancement_proceeds", "float64"),
("repurchase_make_whole_proceeds", "float64"),
("other_foreclosure_proceeds", "float64"),
("non_interest_bearing_upb", "float64"),
("principal_forgiveness_upb", "float64"),
("repurchase_make_whole_proceeds_flag", "category"),
("foreclosure_principal_write_off_amount", "float64"),
("servicing_activity_indicator", "category")
])
print(performance_path)
return cudf.read_csv(performance_path, names=cols, delimiter='|', dtype=list(dtypes.values()), skiprows=1)
def gpu_load_acquisition_csv(acquisition_path, **kwargs):
""" Loads acquisition data
Returns
-------
GPU DataFrame
"""
cols = [
'loan_id', 'orig_channel', 'seller_name', 'orig_interest_rate', 'orig_upb', 'orig_loan_term',
'orig_date', 'first_pay_date', 'orig_ltv', 'orig_cltv', 'num_borrowers', 'dti', 'borrower_credit_score',
'first_home_buyer', 'loan_purpose', 'property_type', 'num_units', 'occupancy_status', 'property_state',
'zip', 'mortgage_insurance_percent', 'product_type', 'coborrow_credit_score', 'mortgage_insurance_type',
'relocation_mortgage_indicator'
]
dtypes = OrderedDict([
("loan_id", "int64"),
("orig_channel", "category"),
("seller_name", "category"),
("orig_interest_rate", "float64"),
("orig_upb", "int64"),
("orig_loan_term", "int64"),
("orig_date", "date"),
("first_pay_date", "date"),
("orig_ltv", "float64"),
("orig_cltv", "float64"),
("num_borrowers", "float64"),
("dti", "float64"),
("borrower_credit_score", "float64"),
("first_home_buyer", "category"),
("loan_purpose", "category"),
("property_type", "category"),
("num_units", "int64"),
("occupancy_status", "category"),
("property_state", "category"),
("zip", "int64"),
("mortgage_insurance_percent", "float64"),
("product_type", "category"),
("coborrow_credit_score", "float64"),
("mortgage_insurance_type", "float64"),
("relocation_mortgage_indicator", "category")
])
print(acquisition_path)
return cudf.read_csv(acquisition_path, names=cols, delimiter='|', dtype=list(dtypes.values()), skiprows=1)
def gpu_load_names(col_path):
""" Loads names used for renaming the banks
Returns
-------
GPU DataFrame
"""
cols = [
'seller_name', 'new'
]
dtypes = OrderedDict([
("seller_name", "category"),
("new", "category"),
])
return cudf.read_csv(col_path, names=cols, delimiter='|', dtype=list(dtypes.values()), skiprows=1)
def create_ever_features(gdf, **kwargs):
everdf = gdf[['loan_id', 'current_loan_delinquency_status']]
everdf = everdf.groupby('loan_id', method='hash').max().reset_index()
del(gdf)
everdf['ever_30'] = (everdf['current_loan_delinquency_status'] >= 1).astype('int8')
everdf['ever_90'] = (everdf['current_loan_delinquency_status'] >= 3).astype('int8')
everdf['ever_180'] = (everdf['current_loan_delinquency_status'] >= 6).astype('int8')
everdf.drop_column('current_loan_delinquency_status')
return everdf
def create_delinq_features(gdf, **kwargs):
delinq_gdf = gdf[['loan_id', 'monthly_reporting_period', 'current_loan_delinquency_status']]
del(gdf)
delinq_30 = delinq_gdf.query('current_loan_delinquency_status >= 1')[['loan_id', 'monthly_reporting_period']].groupby('loan_id', method='hash').min().reset_index()
delinq_30['delinquency_30'] = delinq_30['monthly_reporting_period']
delinq_30.drop_column('monthly_reporting_period')
delinq_90 = delinq_gdf.query('current_loan_delinquency_status >= 3')[['loan_id', 'monthly_reporting_period']].groupby('loan_id', method='hash').min().reset_index()
delinq_90['delinquency_90'] = delinq_90['monthly_reporting_period']
delinq_90.drop_column('monthly_reporting_period')
delinq_180 = delinq_gdf.query('current_loan_delinquency_status >= 6')[['loan_id', 'monthly_reporting_period']].groupby('loan_id', method='hash').min().reset_index()
delinq_180['delinquency_180'] = delinq_180['monthly_reporting_period']
delinq_180.drop_column('monthly_reporting_period')
del(delinq_gdf)
delinq_merge = delinq_30.merge(delinq_90, how='left', on=['loan_id'], type='hash')
delinq_merge['delinquency_90'] = delinq_merge['delinquency_90'].fillna(np.dtype('datetime64[ms]').type('1970-01-01').astype('datetime64[ms]'))
delinq_merge = delinq_merge.merge(delinq_180, how='left', on=['loan_id'], type='hash')
delinq_merge['delinquency_180'] = delinq_merge['delinquency_180'].fillna(np.dtype('datetime64[ms]').type('1970-01-01').astype('datetime64[ms]'))
del(delinq_30)
del(delinq_90)
del(delinq_180)
return delinq_merge
def join_ever_delinq_features(everdf_tmp, delinq_merge, **kwargs):
everdf = everdf_tmp.merge(delinq_merge, on=['loan_id'], how='left', type='hash')
del(everdf_tmp)
del(delinq_merge)
everdf['delinquency_30'] = everdf['delinquency_30'].fillna(np.dtype('datetime64[ms]').type('1970-01-01').astype('datetime64[ms]'))
everdf['delinquency_90'] = everdf['delinquency_90'].fillna(np.dtype('datetime64[ms]').type('1970-01-01').astype('datetime64[ms]'))
everdf['delinquency_180'] = everdf['delinquency_180'].fillna(np.dtype('datetime64[ms]').type('1970-01-01').astype('datetime64[ms]'))
return everdf
def create_joined_df(gdf, everdf, **kwargs):
test = gdf[['loan_id', 'monthly_reporting_period', 'current_loan_delinquency_status', 'current_actual_upb']]
del(gdf)
test['timestamp'] = test['monthly_reporting_period']
test.drop_column('monthly_reporting_period')
test['timestamp_month'] = test['timestamp'].dt.month
test['timestamp_year'] = test['timestamp'].dt.year
test['delinquency_12'] = test['current_loan_delinquency_status']
test.drop_column('current_loan_delinquency_status')
test['upb_12'] = test['current_actual_upb']
test.drop_column('current_actual_upb')
test['upb_12'] = test['upb_12'].fillna(999999999)
test['delinquency_12'] = test['delinquency_12'].fillna(-1)
joined_df = test.merge(everdf, how='left', on=['loan_id'], type='hash')
del(everdf)
del(test)
joined_df['ever_30'] = joined_df['ever_30'].fillna(-1)
joined_df['ever_90'] = joined_df['ever_90'].fillna(-1)
joined_df['ever_180'] = joined_df['ever_180'].fillna(-1)
joined_df['delinquency_30'] = joined_df['delinquency_30'].fillna(-1)
joined_df['delinquency_90'] = joined_df['delinquency_90'].fillna(-1)
joined_df['delinquency_180'] = joined_df['delinquency_180'].fillna(-1)
joined_df['timestamp_year'] = joined_df['timestamp_year'].astype('int32')
joined_df['timestamp_month'] = joined_df['timestamp_month'].astype('int32')
return joined_df
def create_12_mon_features(joined_df, **kwargs):
testdfs = []
n_months = 12
for y in range(1, n_months + 1):
tmpdf = joined_df[['loan_id', 'timestamp_year', 'timestamp_month', 'delinquency_12', 'upb_12']]
tmpdf['josh_months'] = tmpdf['timestamp_year'] * 12 + tmpdf['timestamp_month']
tmpdf['josh_mody_n'] = ((tmpdf['josh_months'].astype('float64') - 24000 - y) / 12).floor()
tmpdf = tmpdf.groupby(['loan_id', 'josh_mody_n'], method='hash').agg({'delinquency_12': 'max','upb_12': 'min'}).reset_index()
tmpdf['delinquency_12'] = (tmpdf['delinquency_12']>3).astype('int32')
tmpdf['delinquency_12'] +=(tmpdf['upb_12']==0).astype('int32')
tmpdf['upb_12'] = tmpdf['upb_12']
tmpdf['timestamp_year'] = (((tmpdf['josh_mody_n'] * n_months) + 24000 + (y - 1)) / 12).floor().astype('int16')
tmpdf['timestamp_month'] = np.int8(y)
tmpdf.drop_column('josh_mody_n')
testdfs.append(tmpdf)
del(tmpdf)
del(joined_df)
return cudf.concat(testdfs)
def combine_joined_12_mon(joined_df, testdf, **kwargs):
joined_df.drop_column('delinquency_12')
joined_df.drop_column('upb_12')
joined_df['timestamp_year'] = joined_df['timestamp_year'].astype('int16')
joined_df['timestamp_month'] = joined_df['timestamp_month'].astype('int8')
return joined_df.merge(testdf, how='left', on=['loan_id', 'timestamp_year', 'timestamp_month'], type='hash')
def final_performance_delinquency(gdf, joined_df, **kwargs):
merged = null_workaround(gdf)
joined_df = null_workaround(joined_df)
merged['timestamp_month'] = merged['monthly_reporting_period'].dt.month
merged['timestamp_month'] = merged['timestamp_month'].astype('int8')
merged['timestamp_year'] = merged['monthly_reporting_period'].dt.year
merged['timestamp_year'] = merged['timestamp_year'].astype('int16')
merged = merged.merge(joined_df, how='left', on=['loan_id', 'timestamp_year', 'timestamp_month'], type='hash')
merged.drop_column('timestamp_year')
merged.drop_column('timestamp_month')
return merged
def join_perf_acq_gdfs(perf, acq, **kwargs):
perf = null_workaround(perf)
acq = null_workaround(acq)
return perf.merge(acq, how='left', on=['loan_id'], type='hash')
def last_mile_cleaning(df, **kwargs):
drop_list = [
'loan_id', 'orig_date', 'first_pay_date', 'seller_name',
'monthly_reporting_period', 'last_paid_installment_date', 'maturity_date', 'ever_30', 'ever_90', 'ever_180',
'delinquency_30', 'delinquency_90', 'delinquency_180', 'upb_12',
'zero_balance_effective_date','foreclosed_after', 'disposition_date','timestamp'
]
for column in drop_list:
df.drop_column(column)
for col, dtype in df.dtypes.iteritems():
if str(dtype)=='category':
df[col] = df[col].cat.codes
df[col] = df[col].astype('float32')
df['delinquency_12'] = df['delinquency_12'] > 0
df['delinquency_12'] = df['delinquency_12'].fillna(False).astype('int32')
for column in df.columns:
df[column] = df[column].fillna(-1)
return df.to_arrow(preserve_index=False)
def main():
parser = argparse.ArgumentParser("rapidssample")
parser.add_argument("--data_dir", type=str, help="location of data")
parser.add_argument("--num_gpu", type=int, help="Number of GPUs to use", default=1)
parser.add_argument("--part_count", type=int, help="Number of data files to train against", default=2)
parser.add_argument("--end_year", type=int, help="Year to end the data load", default=2000)
parser.add_argument("--cpu_predictor", type=str, help="Flag to use CPU for prediction", default='False')
parser.add_argument('-f', type=str, default='') # added for notebook execution scenarios
args = parser.parse_args()
data_dir = args.data_dir
num_gpu = args.num_gpu
part_count = args.part_count
end_year = args.end_year
cpu_predictor = args.cpu_predictor.lower() in ('yes', 'true', 't', 'y', '1')
if cpu_predictor:
print('Training with CPUs require num gpu = 1')
num_gpu = 1
print('data_dir = {0}'.format(data_dir))
print('num_gpu = {0}'.format(num_gpu))
print('part_count = {0}'.format(part_count))
print('end_year = {0}'.format(end_year))
print('cpu_predictor = {0}'.format(cpu_predictor))
import subprocess
cmd = "hostname --all-ip-addresses"
process = subprocess.Popen(cmd.split(), stdout=subprocess.PIPE)
output, error = process.communicate()
IPADDR = str(output.decode()).split()[0]
cluster = LocalCUDACluster(ip=IPADDR,n_workers=num_gpu)
client = Client(cluster)
client
print(client.ncores())
# to download data for this notebook, visit https://rapidsai.github.io/demos/datasets/mortgage-data and update the following paths accordingly
acq_data_path = "{0}/acq".format(data_dir) #"/rapids/data/mortgage/acq"
perf_data_path = "{0}/perf".format(data_dir) #"/rapids/data/mortgage/perf"
col_names_path = "{0}/names.csv".format(data_dir) # "/rapids/data/mortgage/names.csv"
start_year = 2000
client
print('--->>> Workers used: {0}'.format(client.ncores()))
# NOTE: The ETL calculates additional features which are then dropped before creating the XGBoost DMatrix.
# This can be optimized to avoid calculating the dropped features.
print("Reading ...")
t1 = datetime.datetime.now()
gpu_dfs = []
gpu_time = 0
quarter = 1
year = start_year
count = 0
while year <= end_year:
for file in glob(os.path.join(perf_data_path + "/Performance_" + str(year) + "Q" + str(quarter) + "*")):
if count < part_count:
gpu_dfs.append(process_quarter_gpu(client, col_names_path, acq_data_path, year=year, quarter=quarter, perf_file=file))
count += 1
print('file: {0}'.format(file))
print('count: {0}'.format(count))
quarter += 1
if quarter == 5:
year += 1
quarter = 1
wait(gpu_dfs)
t2 = datetime.datetime.now()
print("Reading time: {0}".format(str(t2-t1)))
print('--->>> Number of data parts: {0}'.format(len(gpu_dfs)))
dxgb_gpu_params = {
'nround': 100,
'max_depth': 8,
'max_leaves': 2**8,
'alpha': 0.9,
'eta': 0.1,
'gamma': 0.1,
'learning_rate': 0.1,
'subsample': 1,
'reg_lambda': 1,
'scale_pos_weight': 2,
'min_child_weight': 30,
'tree_method': 'gpu_hist',
'n_gpus': 1,
'distributed_dask': True,
'loss': 'ls',
'objective': 'reg:squarederror',
'max_features': 'auto',
'criterion': 'friedman_mse',
'grow_policy': 'lossguide',
'verbose': True
}
if cpu_predictor:
print('\n---->>>> Training using CPUs <<<<----\n')
dxgb_gpu_params['predictor'] = 'cpu_predictor'
dxgb_gpu_params['tree_method'] = 'hist'
dxgb_gpu_params['objective'] = 'reg:linear'
else:
print('\n---->>>> Training using GPUs <<<<----\n')
print('Training parameters are {0}'.format(dxgb_gpu_params))
gpu_dfs = [delayed(DataFrame.from_arrow)(gpu_df) for gpu_df in gpu_dfs[:part_count]]
gpu_dfs = [gpu_df for gpu_df in gpu_dfs]
wait(gpu_dfs)
tmp_map = [(gpu_df, list(client.who_has(gpu_df).values())[0]) for gpu_df in gpu_dfs]
new_map = {}
for key, value in tmp_map:
if value not in new_map:
new_map[value] = [key]
else:
new_map[value].append(key)
del(tmp_map)
gpu_dfs = []
for list_delayed in new_map.values():
gpu_dfs.append(delayed(cudf.concat)(list_delayed))
del(new_map)
gpu_dfs = [(gpu_df[['delinquency_12']], gpu_df[delayed(list)(gpu_df.columns.difference(['delinquency_12']))]) for gpu_df in gpu_dfs]
gpu_dfs = [(gpu_df[0].persist(), gpu_df[1].persist()) for gpu_df in gpu_dfs]
gpu_dfs = [dask.delayed(xgb.DMatrix)(gpu_df[1], gpu_df[0]) for gpu_df in gpu_dfs]
gpu_dfs = [gpu_df.persist() for gpu_df in gpu_dfs]
gc.collect()
wait(gpu_dfs)
# TRAIN THE MODEL
labels = None
t1 = datetime.datetime.now()
bst = dxgb_gpu.train(client, dxgb_gpu_params, gpu_dfs, labels, num_boost_round=dxgb_gpu_params['nround'])
t2 = datetime.datetime.now()
print('\n---->>>> Training time: {0} <<<<----\n'.format(str(t2-t1)))
print('Exiting script')
if __name__ == '__main__':
main()

621
contrib/fairness/fairlearn-azureml-mitigation.ipynb Normal file
View File

@@ -0,0 +1,621 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copyright (c) Microsoft Corporation. All rights reserved. \n",
"Licensed under the MIT License."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"![Impressions](https://PixelServer20190423114238.azurewebsites.net/api/impressions/MachineLearningNotebooks/contrib/fairness/fairlearn-azureml-mitigation.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Unfairness Mitigation with Fairlearn and Azure Machine Learning\n",
"**This notebook shows how to upload results from Fairlearn's GridSearch mitigation algorithm into a dashboard in Azure Machine Learning Studio**\n",
"\n",
"## Table of Contents\n",
"\n",
"1. [Introduction](#Introduction)\n",
"1. [Loading the Data](#LoadingData)\n",
"1. [Training an Unmitigated Model](#UnmitigatedModel)\n",
"1. [Mitigation with GridSearch](#Mitigation)\n",
"1. [Uploading a Fairness Dashboard to Azure](#AzureUpload)\n",
" 1. Registering models\n",
" 1. Computing Fairness Metrics\n",
" 1. Uploading to Azure\n",
"1. [Conclusion](#Conclusion)\n",
"\n",
"<a id=\"Introduction\"></a>\n",
"## Introduction\n",
"This notebook shows how to use [Fairlearn (an open source fairness assessment and unfairness mitigation package)](http://fairlearn.org) and Azure Machine Learning Studio for a binary classification problem. This example uses the well-known adult census dataset. For the purposes of this notebook, we shall treat this as a loan decision problem. We will pretend that the label indicates whether or not each individual repaid a loan in the past. We will use the data to train a predictor to predict whether previously unseen individuals will repay a loan or not. The assumption is that the model predictions are used to decide whether an individual should be offered a loan. Its purpose is purely illustrative of a workflow including a fairness dashboard - in particular, we do **not** include a full discussion of the detailed issues which arise when considering fairness in machine learning. For such discussions, please [refer to the Fairlearn website](http://fairlearn.org/).\n",
"\n",
"We will apply the [grid search algorithm](https://fairlearn.org/v0.4.6/api_reference/fairlearn.reductions.html#fairlearn.reductions.GridSearch) from the Fairlearn package using a specific notion of fairness called Demographic Parity. This produces a set of models, and we will view these in a dashboard both locally and in the Azure Machine Learning Studio.\n",
"\n",
"### Setup\n",
"\n",
"To use this notebook, an Azure Machine Learning workspace is required.\n",
"Please see the [configuration notebook](../../configuration.ipynb) for information about creating one, if required.\n",
"This notebook also requires the following packages:\n",
"* `azureml-contrib-fairness`\n",
"* `fairlearn>=0.6.2` (pre-v0.5.0 will work with minor modifications)\n",
"* `joblib`\n",
"* `liac-arff`\n",
"* `raiwidgets`\n",
"\n",
"Fairlearn relies on features introduced in v0.22.1 of `scikit-learn`. If you have an older version already installed, please uncomment and run the following cell:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# !pip install --upgrade scikit-learn>=0.22.1"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Finally, please ensure that when you downloaded this notebook, you also downloaded the `fairness_nb_utils.py` file from the same location, and placed it in the same directory as this notebook."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"LoadingData\"></a>\n",
"## Loading the Data\n",
"We use the well-known `adult` census dataset, which we will fetch from the OpenML website. We start with a fairly unremarkable set of imports:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from fairlearn.reductions import GridSearch, DemographicParity, ErrorRate\n",
"from raiwidgets import FairnessDashboard\n",
"\n",
"from sklearn.compose import ColumnTransformer\n",
"from sklearn.impute import SimpleImputer\n",
"from sklearn.linear_model import LogisticRegression\n",
"from sklearn.model_selection import train_test_split\n",
"from sklearn.preprocessing import StandardScaler, OneHotEncoder\n",
"from sklearn.compose import make_column_selector as selector\n",
"from sklearn.pipeline import Pipeline\n",
"\n",
"import pandas as pd"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We can now load and inspect the data:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from fairness_nb_utils import fetch_census_dataset\n",
"\n",
"data = fetch_census_dataset()\n",
" \n",
"# Extract the items we want\n",
"X_raw = data.data\n",
"y = (data.target == '>50K') * 1\n",
"\n",
"X_raw[\"race\"].value_counts().to_dict()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We are going to treat the sex and race of each individual as protected attributes, and in this particular case we are going to remove these attributes from the main data (this is not always the best option - see the [Fairlearn website](http://fairlearn.github.io/) for further discussion). Protected attributes are often denoted by 'A' in the literature, and we follow that convention here:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"A = X_raw[['sex','race']]\n",
"X_raw = X_raw.drop(labels=['sex', 'race'], axis = 1)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We now preprocess our data. To avoid the problem of data leakage, we split our data into training and test sets before performing any other transformations. Subsequent transformations (such as scalings) will be fit to the training data set, and then applied to the test dataset."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"(X_train, X_test, y_train, y_test, A_train, A_test) = train_test_split(\n",
" X_raw, y, A, test_size=0.3, random_state=12345, stratify=y\n",
")\n",
"\n",
"# Ensure indices are aligned between X, y and A,\n",
"# after all the slicing and splitting of DataFrames\n",
"# and Series\n",
"\n",
"X_train = X_train.reset_index(drop=True)\n",
"X_test = X_test.reset_index(drop=True)\n",
"y_train = y_train.reset_index(drop=True)\n",
"y_test = y_test.reset_index(drop=True)\n",
"A_train = A_train.reset_index(drop=True)\n",
"A_test = A_test.reset_index(drop=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We have two types of column in the dataset - categorical columns which will need to be one-hot encoded, and numeric ones which will need to be rescaled. We also need to take care of missing values. We use a simple approach here, but please bear in mind that this is another way that bias could be introduced (especially if one subgroup tends to have more missing values).\n",
"\n",
"For this preprocessing, we make use of `Pipeline` objects from `sklearn`:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"numeric_transformer = Pipeline(\n",
" steps=[\n",
" (\"impute\", SimpleImputer()),\n",
" (\"scaler\", StandardScaler()),\n",
" ]\n",
")\n",
"\n",
"categorical_transformer = Pipeline(\n",
" [\n",
" (\"impute\", SimpleImputer(strategy=\"most_frequent\")),\n",
" (\"ohe\", OneHotEncoder(handle_unknown=\"ignore\", sparse=False)),\n",
" ]\n",
")\n",
"\n",
"preprocessor = ColumnTransformer(\n",
" transformers=[\n",
" (\"num\", numeric_transformer, selector(dtype_exclude=\"category\")),\n",
" (\"cat\", categorical_transformer, selector(dtype_include=\"category\")),\n",
" ]\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now, the preprocessing pipeline is defined, we can run it on our training data, and apply the generated transform to our test data:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"X_train = preprocessor.fit_transform(X_train)\n",
"X_test = preprocessor.transform(X_test)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"UnmitigatedModel\"></a>\n",
"## Training an Unmitigated Model\n",
"\n",
"So we have a point of comparison, we first train a model (specifically, logistic regression from scikit-learn) on the raw data, without applying any mitigation algorithm:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"unmitigated_predictor = LogisticRegression(solver='liblinear', fit_intercept=True)\n",
"\n",
"unmitigated_predictor.fit(X_train, y_train)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We can view this model in the fairness dashboard, and see the disparities which appear:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"FairnessDashboard(sensitive_features=A_test,\n",
" y_true=y_test,\n",
" y_pred={\"unmitigated\": unmitigated_predictor.predict(X_test)})"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Looking at the disparity in accuracy when we select 'Sex' as the sensitive feature, we see that males have an error rate about three times greater than the females. More interesting is the disparity in opportunitiy - males are offered loans at three times the rate of females.\n",
"\n",
"Despite the fact that we removed the feature from the training data, our predictor still discriminates based on sex. This demonstrates that simply ignoring a protected attribute when fitting a predictor rarely eliminates unfairness. There will generally be enough other features correlated with the removed attribute to lead to disparate impact."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"Mitigation\"></a>\n",
"## Mitigation with GridSearch\n",
"\n",
"The `GridSearch` class in `Fairlearn` implements a simplified version of the exponentiated gradient reduction of [Agarwal et al. 2018](https://arxiv.org/abs/1803.02453). The user supplies a standard ML estimator, which is treated as a blackbox - for this simple example, we shall use the logistic regression estimator from scikit-learn. `GridSearch` works by generating a sequence of relabellings and reweightings, and trains a predictor for each.\n",
"\n",
"For this example, we specify demographic parity (on the protected attribute of sex) as the fairness metric. Demographic parity requires that individuals are offered the opportunity (a loan in this example) independent of membership in the protected class (i.e., females and males should be offered loans at the same rate). *We are using this metric for the sake of simplicity* in this example; the appropriate fairness metric can only be selected after *careful examination of the broader context* in which the model is to be used."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"sweep = GridSearch(LogisticRegression(solver='liblinear', fit_intercept=True),\n",
" constraints=DemographicParity(),\n",
" grid_size=71)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"With our estimator created, we can fit it to the data. After `fit()` completes, we extract the full set of predictors from the `GridSearch` object.\n",
"\n",
"The following cell trains a many copies of the underlying estimator, and may take a minute or two to run:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"sweep.fit(X_train, y_train,\n",
" sensitive_features=A_train.sex)\n",
"\n",
"# For Fairlearn pre-v0.5.0, need sweep._predictors\n",
"predictors = sweep.predictors_"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We could load these predictors into the Fairness dashboard now. However, the plot would be somewhat confusing due to their number. In this case, we are going to remove the predictors which are dominated in the error-disparity space by others from the sweep (note that the disparity will only be calculated for the protected attribute; other potentially protected attributes will *not* be mitigated). In general, one might not want to do this, since there may be other considerations beyond the strict optimisation of error and disparity (of the given protected attribute)."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"errors, disparities = [], []\n",
"for predictor in predictors:\n",
" error = ErrorRate()\n",
" error.load_data(X_train, pd.Series(y_train), sensitive_features=A_train.sex)\n",
" disparity = DemographicParity()\n",
" disparity.load_data(X_train, pd.Series(y_train), sensitive_features=A_train.sex)\n",
" \n",
" errors.append(error.gamma(predictor.predict)[0])\n",
" disparities.append(disparity.gamma(predictor.predict).max())\n",
" \n",
"all_results = pd.DataFrame( {\"predictor\": predictors, \"error\": errors, \"disparity\": disparities})\n",
"\n",
"dominant_models_dict = dict()\n",
"base_name_format = \"census_gs_model_{0}\"\n",
"row_id = 0\n",
"for row in all_results.itertuples():\n",
" model_name = base_name_format.format(row_id)\n",
" errors_for_lower_or_eq_disparity = all_results[\"error\"][all_results[\"disparity\"]<=row.disparity]\n",
" if row.error <= errors_for_lower_or_eq_disparity.min():\n",
" dominant_models_dict[model_name] = row.predictor\n",
" row_id = row_id + 1"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We can construct predictions for the dominant models (we include the unmitigated predictor as well, for comparison):"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"predictions_dominant = {\"census_unmitigated\": unmitigated_predictor.predict(X_test)}\n",
"models_dominant = {\"census_unmitigated\": unmitigated_predictor}\n",
"for name, predictor in dominant_models_dict.items():\n",
" value = predictor.predict(X_test)\n",
" predictions_dominant[name] = value\n",
" models_dominant[name] = predictor"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"These predictions may then be viewed in the fairness dashboard. We include the race column from the dataset, as an alternative basis for assessing the models. However, since we have not based our mitigation on it, the variation in the models with respect to race can be large."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"FairnessDashboard(sensitive_features=A_test, \n",
" y_true=y_test.tolist(),\n",
" y_pred=predictions_dominant)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"When using sex as the sensitive feature and accuracy as the metric, we see a Pareto front forming - the set of predictors which represent optimal tradeoffs between accuracy and disparity in predictions. In the ideal case, we would have a predictor at (1,0) - perfectly accurate and without any unfairness under demographic parity (with respect to the protected attribute \"sex\"). The Pareto front represents the closest we can come to this ideal based on our data and choice of estimator. Note the range of the axes - the disparity axis covers more values than the accuracy, so we can reduce disparity substantially for a small loss in accuracy. Finally, we also see that the unmitigated model is towards the top right of the plot, with high accuracy, but worst disparity.\n",
"\n",
"By clicking on individual models on the plot, we can inspect their metrics for disparity and accuracy in greater detail. In a real example, we would then pick the model which represented the best trade-off between accuracy and disparity given the relevant business constraints."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"AzureUpload\"></a>\n",
"## Uploading a Fairness Dashboard to Azure\n",
"\n",
"Uploading a fairness dashboard to Azure is a two stage process. The `FairnessDashboard` invoked in the previous section relies on the underlying Python kernel to compute metrics on demand. This is obviously not available when the fairness dashboard is rendered in AzureML Studio. By default, the dashboard in Azure Machine Learning Studio also requires the models to be registered. The required stages are therefore:\n",
"1. Register the dominant models\n",
"1. Precompute all the required metrics\n",
"1. Upload to Azure\n",
"\n",
"Before that, we need to connect to Azure Machine Learning Studio:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core import Workspace, Experiment, Model\n",
"\n",
"ws = Workspace.from_config()\n",
"ws.get_details()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"RegisterModels\"></a>\n",
"### Registering Models\n",
"\n",
"The fairness dashboard is designed to integrate with registered models, so we need to do this for the models we want in the Studio portal. The assumption is that the names of the models specified in the dashboard dictionary correspond to the `id`s (i.e. `<name>:<version>` pairs) of registered models in the workspace. We register each of the models in the `models_dominant` dictionary into the workspace. For this, we have to save each model to a file, and then register that file:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import joblib\n",
"import os\n",
"\n",
"os.makedirs('models', exist_ok=True)\n",
"def register_model(name, model):\n",
" print(\"Registering \", name)\n",
" model_path = \"models/{0}.pkl\".format(name)\n",
" joblib.dump(value=model, filename=model_path)\n",
" registered_model = Model.register(model_path=model_path,\n",
" model_name=name,\n",
" workspace=ws)\n",
" print(\"Registered \", registered_model.id)\n",
" return registered_model.id\n",
"\n",
"model_name_id_mapping = dict()\n",
"for name, model in models_dominant.items():\n",
" m_id = register_model(name, model)\n",
" model_name_id_mapping[name] = m_id"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now, produce new predictions dictionaries, with the updated names:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"predictions_dominant_ids = dict()\n",
"for name, y_pred in predictions_dominant.items():\n",
" predictions_dominant_ids[model_name_id_mapping[name]] = y_pred"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"PrecomputeMetrics\"></a>\n",
"### Precomputing Metrics\n",
"\n",
"We create a _dashboard dictionary_ using Fairlearn's `metrics` package. The `_create_group_metric_set` method has arguments similar to the Dashboard constructor, except that the sensitive features are passed as a dictionary (to ensure that names are available), and we must specify the type of prediction. Note that we use the `predictions_dominant_ids` dictionary we just created:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"sf = { 'sex': A_test.sex, 'race': A_test.race }\n",
"\n",
"from fairlearn.metrics._group_metric_set import _create_group_metric_set\n",
"\n",
"\n",
"dash_dict = _create_group_metric_set(y_true=y_test,\n",
" predictions=predictions_dominant_ids,\n",
" sensitive_features=sf,\n",
" prediction_type='binary_classification')"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"DashboardUpload\"></a>\n",
"### Uploading the Dashboard\n",
"\n",
"Now, we import our `contrib` package which contains the routine to perform the upload:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.contrib.fairness import upload_dashboard_dictionary, download_dashboard_by_upload_id"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now we can create an Experiment, then a Run, and upload our dashboard to it:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"exp = Experiment(ws, \"Test_Fairlearn_GridSearch_Census_Demo\")\n",
"print(exp)\n",
"\n",
"run = exp.start_logging()\n",
"try:\n",
" dashboard_title = \"Dominant Models from GridSearch\"\n",
" upload_id = upload_dashboard_dictionary(run,\n",
" dash_dict,\n",
" dashboard_name=dashboard_title)\n",
" print(\"\\nUploaded to id: {0}\\n\".format(upload_id))\n",
"\n",
" downloaded_dict = download_dashboard_by_upload_id(run, upload_id)\n",
"finally:\n",
" run.complete()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The dashboard can be viewed in the Run Details page.\n",
"\n",
"Finally, we can verify that the dashboard dictionary which we downloaded matches our upload:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"print(dash_dict == downloaded_dict)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"Conclusion\"></a>\n",
"## Conclusion\n",
"\n",
"In this notebook we have demonstrated how to use the `GridSearch` algorithm from Fairlearn to generate a collection of models, and then present them in the fairness dashboard in Azure Machine Learning Studio. Please remember that this notebook has not attempted to discuss the many considerations which should be part of any approach to unfairness mitigation. The [Fairlearn website](http://fairlearn.org/) provides that discussion"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"authors": [
{
"name": "riedgar"
}
],
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.10"
}
},
"nbformat": 4,
"nbformat_minor": 2
}

11
contrib/fairness/fairlearn-azureml-mitigation.yml Normal file
View File

@@ -0,0 +1,11 @@
name: fairlearn-azureml-mitigation
dependencies:
- pip:
- azureml-sdk
- azureml-contrib-fairness
- fairlearn>=0.6.2
- joblib
- liac-arff
- raiwidgets~=0.17.0
- itsdangerous==2.0.1
- markupsafe<2.1.0

111
contrib/fairness/fairness_nb_utils.py Normal file
View File

@@ -0,0 +1,111 @@
# ---------------------------------------------------------
# Copyright (c) Microsoft Corporation. All rights reserved.
# ---------------------------------------------------------
"""Utilities for azureml-contrib-fairness notebooks."""
import arff
from collections import OrderedDict
from contextlib import closing
import gzip
import pandas as pd
from sklearn.datasets import fetch_openml
from sklearn.utils import Bunch
import time
def fetch_openml_with_retries(data_id, max_retries=4, retry_delay=60):
"""Fetch a given dataset from OpenML with retries as specified."""
for i in range(max_retries):
try:
print("Download attempt {0} of {1}".format(i + 1, max_retries))
data = fetch_openml(data_id=data_id, as_frame=True)
break
except Exception as e: # noqa: B902
print("Download attempt failed with exception:")
print(e)
if i + 1 != max_retries:
print("Will retry after {0} seconds".format(retry_delay))
time.sleep(retry_delay)
retry_delay = retry_delay * 2
else:
raise RuntimeError("Unable to download dataset from OpenML")
return data
_categorical_columns = [
'workclass',
'education',
'marital-status',
'occupation',
'relationship',
'race',
'sex',
'native-country'
]
def fetch_census_dataset():
"""Fetch the Adult Census Dataset.
This uses a particular URL for the Adult Census dataset. The code
is a simplified version of fetch_openml() in sklearn.
The data are copied from:
https://openml.org/data/v1/download/1595261.gz
(as of 2021-03-31)
"""
try:
from urllib import urlretrieve
except ImportError:
from urllib.request import urlretrieve
filename = "1595261.gz"
data_url = "https://rainotebookscdn.blob.core.windows.net/datasets/"
remaining_attempts = 5
sleep_duration = 10
while remaining_attempts > 0:
try:
urlretrieve(data_url + filename, filename)
http_stream = gzip.GzipFile(filename=filename, mode='rb')
with closing(http_stream):
def _stream_generator(response):
for line in response:
yield line.decode('utf-8')
stream = _stream_generator(http_stream)
data = arff.load(stream)
except Exception as exc: # noqa: B902
remaining_attempts -= 1
print("Error downloading dataset from {} ({} attempt(s) remaining)"
.format(data_url, remaining_attempts))
print(exc)
time.sleep(sleep_duration)
sleep_duration *= 2
continue
else:
# dataset successfully downloaded
break
else:
raise Exception("Could not retrieve dataset from {}.".format(data_url))
attributes = OrderedDict(data['attributes'])
arff_columns = list(attributes)
raw_df = pd.DataFrame(data=data['data'], columns=arff_columns)
target_column_name = 'class'
target = raw_df.pop(target_column_name)
for col_name in _categorical_columns:
dtype = pd.api.types.CategoricalDtype(attributes[col_name])
raw_df[col_name] = raw_df[col_name].astype(dtype, copy=False)
result = Bunch()
result.data = raw_df
result.target = target
return result

545
contrib/fairness/upload-fairness-dashboard.ipynb Normal file
View File

@@ -0,0 +1,545 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copyright (c) Microsoft Corporation. All rights reserved. \n",
"Licensed under the MIT License."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"![Impressions](https://PixelServer20190423114238.azurewebsites.net/api/impressions/MachineLearningNotebooks/contrib/fairness/upload-fairness-dashboard.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Upload a Fairness Dashboard to Azure Machine Learning Studio\n",
"**This notebook shows how to generate and upload a fairness assessment dashboard from Fairlearn to AzureML Studio**\n",
"\n",
"## Table of Contents\n",
"\n",
"1. [Introduction](#Introduction)\n",
"1. [Loading the Data](#LoadingData)\n",
"1. [Processing the Data](#ProcessingData)\n",
"1. [Training Models](#TrainingModels)\n",
"1. [Logging in to AzureML](#LoginAzureML)\n",
"1. [Registering the Models](#RegisterModels)\n",
"1. [Using the Fairness Dashboard](#LocalDashboard)\n",
"1. [Uploading a Fairness Dashboard to Azure](#AzureUpload)\n",
" 1. Computing Fairness Metrics\n",
" 1. Uploading to Azure\n",
"1. [Conclusion](#Conclusion)\n",
" \n",
"\n",
"<a id=\"Introduction\"></a>\n",
"## Introduction\n",
"\n",
"In this notebook, we walk through a simple example of using the `azureml-contrib-fairness` package to upload a collection of fairness statistics for a fairness dashboard. It is an example of integrating the [open source Fairlearn package](https://www.github.com/fairlearn/fairlearn) with Azure Machine Learning. This is not an example of fairness analysis or mitigation - this notebook simply shows how to get a fairness dashboard into the Azure Machine Learning portal. We will load the data and train a couple of simple models. We will then use Fairlearn to generate data for a Fairness dashboard, which we can upload to Azure Machine Learning portal and view there.\n",
"\n",
"### Setup\n",
"\n",
"To use this notebook, an Azure Machine Learning workspace is required.\n",
"Please see the [configuration notebook](../../configuration.ipynb) for information about creating one, if required.\n",
"This notebook also requires the following packages:\n",
"* `azureml-contrib-fairness`\n",
"* `fairlearn>=0.6.2` (also works for pre-v0.5.0 with slight modifications)\n",
"* `joblib`\n",
"* `liac-arff`\n",
"* `raiwidgets`\n",
"\n",
"Fairlearn relies on features introduced in v0.22.1 of `scikit-learn`. If you have an older version already installed, please uncomment and run the following cell:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# !pip install --upgrade scikit-learn>=0.22.1"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Finally, please ensure that when you downloaded this notebook, you also downloaded the `fairness_nb_utils.py` file from the same location, and placed it in the same directory as this notebook."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"LoadingData\"></a>\n",
"## Loading the Data\n",
"We use the well-known `adult` census dataset, which we fetch from the OpenML website. We start with a fairly unremarkable set of imports:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from sklearn import svm\n",
"from sklearn.compose import ColumnTransformer\n",
"from sklearn.impute import SimpleImputer\n",
"from sklearn.linear_model import LogisticRegression\n",
"from sklearn.model_selection import train_test_split\n",
"from sklearn.preprocessing import StandardScaler, OneHotEncoder\n",
"from sklearn.compose import make_column_selector as selector\n",
"from sklearn.pipeline import Pipeline"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now we can load the data:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from fairness_nb_utils import fetch_census_dataset\n",
"\n",
"data = fetch_census_dataset()\n",
" \n",
"# Extract the items we want\n",
"X_raw = data.data\n",
"y = (data.target == '>50K') * 1"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We can take a look at some of the data. For example, the next cells shows the counts of the different races identified in the dataset:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"print(X_raw[\"race\"].value_counts().to_dict())"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"ProcessingData\"></a>\n",
"## Processing the Data\n",
"\n",
"With the data loaded, we process it for our needs. First, we extract the sensitive features of interest into `A` (conventionally used in the literature) and leave the rest of the feature data in `X_raw`:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"A = X_raw[['sex','race']]\n",
"X_raw = X_raw.drop(labels=['sex', 'race'],axis = 1)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We now preprocess our data. To avoid the problem of data leakage, we split our data into training and test sets before performing any other transformations. Subsequent transformations (such as scalings) will be fit to the training data set, and then applied to the test dataset."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"(X_train, X_test, y_train, y_test, A_train, A_test) = train_test_split(\n",
" X_raw, y, A, test_size=0.3, random_state=12345, stratify=y\n",
")\n",
"\n",
"# Ensure indices are aligned between X, y and A,\n",
"# after all the slicing and splitting of DataFrames\n",
"# and Series\n",
"\n",
"X_train = X_train.reset_index(drop=True)\n",
"X_test = X_test.reset_index(drop=True)\n",
"y_train = y_train.reset_index(drop=True)\n",
"y_test = y_test.reset_index(drop=True)\n",
"A_train = A_train.reset_index(drop=True)\n",
"A_test = A_test.reset_index(drop=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We have two types of column in the dataset - categorical columns which will need to be one-hot encoded, and numeric ones which will need to be rescaled. We also need to take care of missing values. We use a simple approach here, but please bear in mind that this is another way that bias could be introduced (especially if one subgroup tends to have more missing values).\n",
"\n",
"For this preprocessing, we make use of `Pipeline` objects from `sklearn`:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"numeric_transformer = Pipeline(\n",
" steps=[\n",
" (\"impute\", SimpleImputer()),\n",
" (\"scaler\", StandardScaler()),\n",
" ]\n",
")\n",
"\n",
"categorical_transformer = Pipeline(\n",
" [\n",
" (\"impute\", SimpleImputer(strategy=\"most_frequent\")),\n",
" (\"ohe\", OneHotEncoder(handle_unknown=\"ignore\", sparse=False)),\n",
" ]\n",
")\n",
"\n",
"preprocessor = ColumnTransformer(\n",
" transformers=[\n",
" (\"num\", numeric_transformer, selector(dtype_exclude=\"category\")),\n",
" (\"cat\", categorical_transformer, selector(dtype_include=\"category\")),\n",
" ]\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now, the preprocessing pipeline is defined, we can run it on our training data, and apply the generated transform to our test data:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"X_train = preprocessor.fit_transform(X_train)\n",
"X_test = preprocessor.transform(X_test)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"TrainingModels\"></a>\n",
"## Training Models\n",
"\n",
"We now train a couple of different models on our data. The `adult` census dataset is a classification problem - the goal is to predict whether a particular individual exceeds an income threshold. For the purpose of generating a dashboard to upload, it is sufficient to train two basic classifiers. First, a logistic regression classifier:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"lr_predictor = LogisticRegression(solver='liblinear', fit_intercept=True)\n",
"\n",
"lr_predictor.fit(X_train, y_train)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"And for comparison, a support vector classifier:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"svm_predictor = svm.SVC()\n",
"\n",
"svm_predictor.fit(X_train, y_train)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"LoginAzureML\"></a>\n",
"## Logging in to AzureML\n",
"\n",
"With our two classifiers trained, we can log into our AzureML workspace:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core import Workspace, Experiment, Model\n",
"\n",
"ws = Workspace.from_config()\n",
"ws.get_details()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"RegisterModels\"></a>\n",
"## Registering the Models\n",
"\n",
"Next, we register our models. By default, the subroutine which uploads the models checks that the names provided correspond to registered models in the workspace. We define a utility routine to do the registering:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import joblib\n",
"import os\n",
"\n",
"os.makedirs('models', exist_ok=True)\n",
"def register_model(name, model):\n",
" print(\"Registering \", name)\n",
" model_path = \"models/{0}.pkl\".format(name)\n",
" joblib.dump(value=model, filename=model_path)\n",
" registered_model = Model.register(model_path=model_path,\n",
" model_name=name,\n",
" workspace=ws)\n",
" print(\"Registered \", registered_model.id)\n",
" return registered_model.id"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now, we register the models. For convenience in subsequent method calls, we store the results in a dictionary, which maps the `id` of the registered model (a string in `name:version` format) to the predictor itself:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"model_dict = {}\n",
"\n",
"lr_reg_id = register_model(\"fairness_linear_regression\", lr_predictor)\n",
"model_dict[lr_reg_id] = lr_predictor\n",
"svm_reg_id = register_model(\"fairness_svm\", svm_predictor)\n",
"model_dict[svm_reg_id] = svm_predictor"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"LocalDashboard\"></a>\n",
"## Using the Fairlearn Dashboard\n",
"\n",
"We can now examine the fairness of the two models we have training, both as a function of race and (binary) sex. Before uploading the dashboard to the AzureML portal, we will first instantiate a local instance of the Fairlearn dashboard.\n",
"\n",
"Regardless of the viewing location, the dashboard is based on three things - the true values, the model predictions and the sensitive feature values. The dashboard can use predictions from multiple models and multiple sensitive features if desired (as we are doing here).\n",
"\n",
"Our first step is to generate a dictionary mapping the `id` of the registered model to the corresponding array of predictions:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"ys_pred = {}\n",
"for n, p in model_dict.items():\n",
" ys_pred[n] = p.predict(X_test)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We can examine these predictions in a locally invoked Fairlearn dashboard. This can be compared to the dashboard uploaded to the portal (in the next section):"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from raiwidgets import FairnessDashboard\n",
"\n",
"FairnessDashboard(sensitive_features=A_test, \n",
" y_true=y_test.tolist(),\n",
" y_pred=ys_pred)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"AzureUpload\"></a>\n",
"## Uploading a Fairness Dashboard to Azure\n",
"\n",
"Uploading a fairness dashboard to Azure is a two stage process. The `FairnessDashboard` invoked in the previous section relies on the underlying Python kernel to compute metrics on demand. This is obviously not available when the fairness dashboard is rendered in AzureML Studio. The required stages are therefore:\n",
"1. Precompute all the required metrics\n",
"1. Upload to Azure\n",
"\n",
"\n",
"### Computing Fairness Metrics\n",
"We use Fairlearn to create a dictionary which contains all the data required to display a dashboard. This includes both the raw data (true values, predicted values and sensitive features), and also the fairness metrics. The API is similar to that used to invoke the Dashboard locally. However, there are a few minor changes to the API, and the type of problem being examined (binary classification, regression etc.) needs to be specified explicitly:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"sf = { 'Race': A_test.race, 'Sex': A_test.sex }\n",
"\n",
"from fairlearn.metrics._group_metric_set import _create_group_metric_set\n",
"\n",
"dash_dict = _create_group_metric_set(y_true=y_test,\n",
" predictions=ys_pred,\n",
" sensitive_features=sf,\n",
" prediction_type='binary_classification')"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The `_create_group_metric_set()` method is currently underscored since its exact design is not yet final in Fairlearn."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Uploading to Azure\n",
"\n",
"We can now import the `azureml.contrib.fairness` package itself. We will round-trip the data, so there are two required subroutines:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.contrib.fairness import upload_dashboard_dictionary, download_dashboard_by_upload_id"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Finally, we can upload the generated dictionary to AzureML. The upload method requires a run, so we first create an experiment and a run. The uploaded dashboard can be seen on the corresponding Run Details page in AzureML Studio. For completeness, we also download the dashboard dictionary which we uploaded."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"exp = Experiment(ws, \"notebook-01\")\n",
"print(exp)\n",
"\n",
"run = exp.start_logging()\n",
"try:\n",
" dashboard_title = \"Sample notebook upload\"\n",
" upload_id = upload_dashboard_dictionary(run,\n",
" dash_dict,\n",
" dashboard_name=dashboard_title)\n",
" print(\"\\nUploaded to id: {0}\\n\".format(upload_id))\n",
"\n",
" downloaded_dict = download_dashboard_by_upload_id(run, upload_id)\n",
"finally:\n",
" run.complete()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Finally, we can verify that the dashboard dictionary which we downloaded matches our upload:"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"print(dash_dict == downloaded_dict)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"<a id=\"Conclusion\"></a>\n",
"## Conclusion\n",
"\n",
"In this notebook we have demonstrated how to generate and upload a fairness dashboard to AzureML Studio. We have not discussed how to analyse the results and apply mitigations. Those topics will be covered elsewhere."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"authors": [
{
"name": "riedgar"
}
],
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.10"
}
},
"nbformat": 4,
"nbformat_minor": 4
}

11
contrib/fairness/upload-fairness-dashboard.yml Normal file
View File

@@ -0,0 +1,11 @@
name: upload-fairness-dashboard
dependencies:
- pip:
- azureml-sdk
- azureml-contrib-fairness
- fairlearn>=0.6.2
- joblib
- liac-arff
- raiwidgets~=0.17.0
- itsdangerous==2.0.1
- markupsafe<2.1.0

15
how-to-use-azureml/README.md Normal file
View File

@@ -0,0 +1,15 @@
## Examples to get started with Azure Machine Learning service
Learn how to use Azure Machine Learning services for experimentation and model management.
As a pre-requisite, run the [configuration Notebook](../configuration.ipynb) notebook first to set up your Azure ML Workspace. Then, run the notebooks in following recommended order.
* [train-within-notebook](./training/train-within-notebook): Train a model while tracking run history, and learn how to deploy the model as web service to Azure Container Instance.
* [train-on-local](./training/train-on-local): Learn how to submit a run to local computer and use Azure ML managed run configuration.
* [train-on-amlcompute](./training/train-on-amlcompute): Use a 1-n node Azure ML managed compute cluster for remote runs on Azure CPU or GPU infrastructure.
* [train-on-remote-vm](./training/train-on-remote-vm): Use Data Science Virtual Machine as a target for remote runs.
* [logging-api](./track-and-monitor-experiments/logging-api): Learn about the details of logging metrics to run history.
* [production-deploy-to-aks](./deployment/production-deploy-to-aks) Deploy a model to production at scale on Azure Kubernetes Service.
* [enable-app-insights-in-production-service](./deployment/enable-app-insights-in-production-service) Learn how to use App Insights with production web service.
Find quickstarts, end-to-end tutorials, and how-tos on the [official documentation site for Azure Machine Learning service](https://docs.microsoft.com/en-us/azure/machine-learning/service/).

299
how-to-use-azureml/automated-machine-learning/README.md Normal file
View File

@@ -0,0 +1,299 @@
# Table of Contents
1. [Automated ML Introduction](#introduction)
1. [Setup using Compute Instances](#jupyter)
1. [Setup using a Local Conda environment](#localconda)
1. [Setup using Azure Databricks](#databricks)
1. [Automated ML SDK Sample Notebooks](#samples)
1. [Documentation](#documentation)
1. [Running using python command](#pythoncommand)
1. [Troubleshooting](#troubleshooting)
<a name="introduction"></a>
# Automated ML introduction
Automated machine learning (automated ML) builds high quality machine learning models for you by automating model and hyperparameter selection. Bring a labelled dataset that you want to build a model for, automated ML will give you a high quality machine learning model that you can use for predictions.
If you are new to Data Science, automated ML will help you get jumpstarted by simplifying machine learning model building. It abstracts you from needing to perform model selection, hyperparameter selection and in one step creates a high quality trained model for you to use.
If you are an experienced data scientist, automated ML will help increase your productivity by intelligently performing the model and hyperparameter selection for your training and generates high quality models much quicker than manually specifying several combinations of the parameters and running training jobs. Automated ML provides visibility and access to all the training jobs and the performance characteristics of the models to help you further tune the pipeline if you desire.
Below are the three execution environments supported by automated ML.
<a name="jupyter"></a>
## Setup using Compute Instances - Jupyter based notebooks from a Azure Virtual Machine
1. Open the [ML Azure portal](https://ml.azure.com)
1. Select Compute
1. Select Compute Instances
1. Click New
1. Type a Compute Name, select a Virtual Machine type and select a Virtual Machine size
1. Click Create
<a name="localconda"></a>
## Setup using a Local Conda environment
To run these notebook on your own notebook server, use these installation instructions.
The instructions below will install everything you need and then start a Jupyter notebook.
### 1. Install mini-conda from [here](https://conda.io/miniconda.html), choose 64-bit Python 3.7 or higher.
- **Note**: if you already have conda installed, you can keep using it but it should be version 4.4.10 or later (as shown by: conda -V). If you have a previous version installed, you can update it using the command: conda update conda.
There's no need to install mini-conda specifically.
### 2. Downloading the sample notebooks
- Download the sample notebooks from [GitHub](https://github.com/Azure/MachineLearningNotebooks) as zip and extract the contents to a local directory. The automated ML sample notebooks are in the "automated-machine-learning" folder.
### 3. Setup a new conda environment
The **automl_setup** script creates a new conda environment, installs the necessary packages, configures the widget and starts a jupyter notebook. It takes the conda environment name as an optional parameter. The default conda environment name is azure_automl. The exact command depends on the operating system. See the specific sections below for Windows, Mac and Linux. It can take about 10 minutes to execute.
Packages installed by the **automl_setup** script:
<ul><li>python</li><li>nb_conda</li><li>matplotlib</li><li>numpy</li><li>cython</li><li>urllib3</li><li>scipy</li><li>scikit-learn</li><li>pandas</li><li>tensorflow</li><li>py-xgboost</li><li>azureml-sdk</li><li>azureml-widgets</li><li>pandas-ml</li></ul>
For more details refer to the [automl_env.yml](./automl_env.yml)
## Windows
Start an **Anaconda Prompt** window, cd to the **how-to-use-azureml/automated-machine-learning** folder where the sample notebooks were extracted and then run:
```
automl_setup
```
## Mac
Install "Command line developer tools" if it is not already installed (you can use the command: `xcode-select --install`).
Start a Terminal windows, cd to the **how-to-use-azureml/automated-machine-learning** folder where the sample notebooks were extracted and then run:
```
bash automl_setup_mac.sh
```
## Linux
cd to the **how-to-use-azureml/automated-machine-learning** folder where the sample notebooks were extracted and then run:
```
bash automl_setup_linux.sh
```
### 4. Running configuration.ipynb
- Before running any samples you next need to run the configuration notebook. Click on [configuration](../../configuration.ipynb) notebook
- Execute the cells in the notebook to Register Machine Learning Services Resource Provider and create a workspace. (*instructions in notebook*)
### 5. Running Samples
- Please make sure you use the Python [conda env:azure_automl] kernel when trying the sample Notebooks.
- Follow the instructions in the individual notebooks to explore various features in automated ML.
### 6. Starting jupyter notebook manually
To start your Jupyter notebook manually, use:
```
conda activate azure_automl
jupyter notebook
```
or on Mac or Linux:
```
source activate azure_automl
jupyter notebook
```
<a name="databricks"></a>
## Setup using Azure Databricks
**NOTE**: Please create your Azure Databricks cluster as v7.1 (high concurrency preferred) with **Python 3** (dropdown).
**NOTE**: You should at least have contributor access to your Azure subcription to run the notebook.
- You can find the detail Readme instructions at [GitHub](https://github.com/Azure/MachineLearningNotebooks/tree/master/how-to-use-azureml/azure-databricks/automl).
- Download the sample notebook automl-databricks-local-01.ipynb from [GitHub](https://github.com/Azure/MachineLearningNotebooks/tree/master/how-to-use-azureml/azure-databricks/automl) and import into the Azure databricks workspace.
- Attach the notebook to the cluster.
<a name="samples"></a>
# Automated ML SDK Sample Notebooks
## Classification
- **Classify Credit Card Fraud**
- Dataset: [Kaggle's credit card fraud detection dataset](https://www.kaggle.com/mlg-ulb/creditcardfraud)
- **[Jupyter Notebook (remote run)](classification-credit-card-fraud/auto-ml-classification-credit-card-fraud.ipynb)**
- run the experiment remotely on AML Compute cluster
- test the performance of the best model in the local environment
- **[Jupyter Notebook (local run)](local-run-classification-credit-card-fraud/auto-ml-classification-credit-card-fraud-local.ipynb)**
- run experiment in the local environment
- use Mimic Explainer for computing feature importance
- deploy the best model along with the explainer to an Azure Kubernetes (AKS) cluster, which will compute the raw and engineered feature importances at inference time
- **Predict Term Deposit Subscriptions in a Bank**
- Dataset: [UCI's bank marketing dataset](https://www.kaggle.com/janiobachmann/bank-marketing-dataset)
- **[Jupyter Notebook](classification-bank-marketing-all-features/auto-ml-classification-bank-marketing-all-features.ipynb)**
- run experiment remotely on AML Compute cluster to generate ONNX compatible models
- view the featurization steps that were applied during training
- view feature importance for the best model
- download the best model in ONNX format and use it for inferencing using ONNXRuntime
- deploy the best model in PKL format to Azure Container Instance (ACI)
- **Predict Newsgroup based on Text from News Article**
- Dataset: [20 newsgroups text dataset](https://scikit-learn.org/0.19/datasets/twenty_newsgroups.html)
- **[Jupyter Notebook](classification-text-dnn/auto-ml-classification-text-dnn.ipynb)**
- AutoML highlights here include using deep neural networks (DNNs) to create embedded features from text data
- AutoML will use Bidirectional Encoder Representations from Transformers (BERT) when a GPU compute is used
- Bidirectional Long-Short Term neural network (BiLSTM) will be utilized when a CPU compute is used, thereby optimizing the choice of DNN
## Regression
- **Predict Performance of Hardware Parts**
- Dataset: Hardware Performance Dataset
- **[Jupyter Notebook](regression/auto-ml-regression.ipynb)**
- run the experiment remotely on AML Compute cluster
- get best trained model for a different metric than the one the experiment was optimized for
- test the performance of the best model in the local environment
- **[Jupyter Notebook (advanced)](regression/auto-ml-regression.ipynb)**
- run the experiment remotely on AML Compute cluster
- customize featurization: override column purpose within the dataset, configure transformer parameters
- get best trained model for a different metric than the one the experiment was optimized for
- run a model explanation experiment on the remote cluster
- deploy the model along the explainer and run online inferencing
## Time Series Forecasting
- **Forecast Energy Demand**
- Dataset: [NYC energy demand data](http://mis.nyiso.com/public/P-58Blist.htm)
- **[Jupyter Notebook](forecasting-energy-demand/auto-ml-forecasting-energy-demand.ipynb)**
- run experiment remotely on AML Compute cluster
- use lags and rolling window features
- view the featurization steps that were applied during training
- get the best model, use it to forecast on test data and compare the accuracy of predictions against real data
- **Forecast Orange Juice Sales (Multi-Series)**
- Dataset: [Dominick's grocery sales of orange juice](forecasting-orange-juice-sales/dominicks_OJ.csv)
- **[Jupyter Notebook](forecasting-orange-juice-sales/dominicks_OJ.csv)**
- run experiment remotely on AML Compute cluster
- customize time-series featurization, change column purpose and override transformer hyper parameters
- evaluate locally the performance of the generated best model
- deploy the best model as a webservice on Azure Container Instance (ACI)
- get online predictions from the deployed model
- **Forecast Demand of a Bike-Sharing Service**
- Dataset: [Bike demand data](forecasting-bike-share/bike-no.csv)
- **[Jupyter Notebook](forecasting-bike-share/auto-ml-forecasting-bike-share.ipynb)**
- run experiment remotely on AML Compute cluster
- integrate holiday features
- run rolling forecast for test set that is longer than the forecast horizon
- compute metrics on the predictions from the remote forecast
- **The Forecast Function Interface**
- Dataset: Generated for sample purposes
- **[Jupyter Notebook](forecasting-forecast-function/auto-ml-forecasting-function.ipynb)**
- train a forecaster using a remote AML Compute cluster
- capabilities of forecast function (e.g. forecast farther into the horizon)
- generate confidence intervals
- **Forecast Beverage Production**
- Dataset: [Monthly beer production data](forecasting-beer-remote/Beer_no_valid_split_train.csv)
- **[Jupyter Notebook](forecasting-beer-remote/auto-ml-forecasting-beer-remote.ipynb)**
- train using a remote AML Compute cluster
- enable the DNN learning model
- forecast on a remote compute cluster and compare different model performance
- **Continuous Retraining with NOAA Weather Data**
- Dataset: [NOAA weather data from Azure Open Datasets](https://azure.microsoft.com/en-us/services/open-datasets/)
- **[Jupyter Notebook](continuous-retraining/auto-ml-continuous-retraining.ipynb)**
- continuously retrain a model using Pipelines and AutoML
- create a Pipeline to upload a time series dataset to an Azure blob
- create a Pipeline to run an AutoML experiment and register the best resulting model in the Workspace
- publish the training pipeline created and schedule it to run daily
<a name="documentation"></a>
See [Configure automated machine learning experiments](https://docs.microsoft.com/azure/machine-learning/service/how-to-configure-auto-train) to learn how more about the the settings and features available for automated machine learning experiments.
<a name="pythoncommand"></a>
# Running using python command
Jupyter notebook provides a File / Download as / Python (.py) option for saving the notebook as a Python file.
You can then run this file using the python command.
However, on Windows the file needs to be modified before it can be run.
The following condition must be added to the main code in the file:
if __name__ == "__main__":
The main code of the file must be indented so that it is under this condition.
<a name="troubleshooting"></a>
# Troubleshooting
## automl_setup fails
1. On Windows, make sure that you are running automl_setup from an Anconda Prompt window rather than a regular cmd window. You can launch the "Anaconda Prompt" window by hitting the Start button and typing "Anaconda Prompt". If you don't see the application "Anaconda Prompt", you might not have conda or mini conda installed. In that case, you can install it [here](https://conda.io/miniconda.html)
2. Check that you have conda 64-bit installed rather than 32-bit. You can check this with the command `conda info`. The `platform` should be `win-64` for Windows or `osx-64` for Mac.
3. Check that you have conda 4.7.8 or later. You can check the version with the command `conda -V`. If you have a previous version installed, you can update it using the command: `conda update conda`.
4. On Linux, if the error is `gcc: error trying to exec 'cc1plus': execvp: No such file or directory`, install build essentials using the command `sudo apt-get install build-essential`.
5. Pass a new name as the first parameter to automl_setup so that it creates a new conda environment. You can view existing conda environments using `conda env list` and remove them with `conda env remove -n <environmentname>`.
## automl_setup_linux.sh fails
If automl_setup_linux.sh fails on Ubuntu Linux with the error: `unable to execute 'gcc': No such file or directory`
1. Make sure that outbound ports 53 and 80 are enabled. On an Azure VM, you can do this from the Azure Portal by selecting the VM and clicking on Networking.
2. Run the command: `sudo apt-get update`
3. Run the command: `sudo apt-get install build-essential --fix-missing`
4. Run `automl_setup_linux.sh` again.
## configuration.ipynb fails
1) For local conda, make sure that you have susccessfully run automl_setup first.
2) Check that the subscription_id is correct. You can find the subscription_id in the Azure Portal by selecting All Service and then Subscriptions. The characters "<" and ">" should not be included in the subscription_id value. For example, `subscription_id = "12345678-90ab-1234-5678-1234567890abcd"` has the valid format.
3) Check that you have Contributor or Owner access to the Subscription.
4) Check that the region is one of the supported regions: `eastus2`, `eastus`, `westcentralus`, `southeastasia`, `westeurope`, `australiaeast`, `westus2`, `southcentralus`
5) Check that you have access to the region using the Azure Portal.
## import AutoMLConfig fails after upgrade from before 1.0.76 to 1.0.76 or later
There were package changes in automated machine learning version 1.0.76, which require the previous version to be uninstalled before upgrading to the new version.
If you have manually upgraded from a version of automated machine learning before 1.0.76 to 1.0.76 or later, you may get the error:
`ImportError: cannot import name 'AutoMLConfig'`
This can be resolved by running:
`pip uninstall azureml-train-automl` and then
`pip install azureml-train-automl`
The automl_setup.cmd script does this automatically.
## workspace.from_config fails
If the call `ws = Workspace.from_config()` fails:
1) Make sure that you have run the `configuration.ipynb` notebook successfully.
2) If you are running a notebook from a folder that is not under the folder where you ran `configuration.ipynb`, copy the folder aml_config and the file config.json that it contains to the new folder. Workspace.from_config reads the config.json for the notebook folder or it parent folder.
3) If you are switching to a new subscription, resource group, workspace or region, make sure that you run the `configuration.ipynb` notebook again. Changing config.json directly will only work if the workspace already exists in the specified resource group under the specified subscription.
4) If you want to change the region, please change the workspace, resource group or subscription. `Workspace.create` will not create or update a workspace if it already exists, even if the region specified is different.
## Sample notebook fails
If a sample notebook fails with an error that property, method or library does not exist:
1) Check that you have selected correct kernel in jupyter notebook. The kernel is displayed in the top right of the notebook page. It can be changed using the `Kernel | Change Kernel` menu option. For Azure Notebooks, it should be `Python 3.6`. For local conda environments, it should be the conda envioronment name that you specified in automl_setup. The default is azure_automl. Note that the kernel is saved as part of the notebook. So, if you switch to a new conda environment, you will have to select the new kernel in the notebook.
2) Check that the notebook is for the SDK version that you are using. You can check the SDK version by executing `azureml.core.VERSION` in a jupyter notebook cell. You can download previous version of the sample notebooks from GitHub by clicking the `Branch` button, selecting the `Tags` tab and then selecting the version.
## Numpy import fails on Windows
Some Windows environments see an error loading numpy with the latest Python version 3.6.8. If you see this issue, try with Python version 3.6.7.
## Numpy import fails
Check the tensorflow version in the automated ml conda environment. Supported versions are < 1.13. Uninstall tensorflow from the environment if version is >= 1.13
You may check the version of tensorflow and uninstall as follows
1) start a command shell, activate conda environment where automated ml packages are installed
2) enter `pip freeze` and look for `tensorflow` , if found, the version listed should be < 1.13
3) If the listed version is a not a supported version, `pip uninstall tensorflow` in the command shell and enter y for confirmation.
## KeyError: 'brand' when running AutoML on local compute or Azure Databricks cluster**
If a new environment was created after 10 June 2020 using SDK 1.7.0 or lower, training may fail with the above error due to an update in the py-cpuinfo package. (Environments created on or before 10 June 2020 are unaffected, as well as experiments run on remote compute as cached training images are used.) To work around this issue, either of the two following steps can be taken:
1) Update the SDK version to 1.8.0 or higher (this will also downgrade py-cpuinfo to 5.0.0):
`pip install --upgrade azureml-sdk[automl]`
2) Downgrade the installed version of py-cpuinfo to 5.0.0:
`pip install py-cpuinfo==5.0.0`
## Remote run: DsvmCompute.create fails
There are several reasons why the DsvmCompute.create can fail. The reason is usually in the error message but you have to look at the end of the error message for the detailed reason. Some common reasons are:
1) `Compute name is invalid, it should start with a letter, be between 2 and 16 character, and only include letters (a-zA-Z), numbers (0-9) and \'-\'.` Note that underscore is not allowed in the name.
2) `The requested VM size xxxxx is not available in the current region.` You can select a different region or vm_size.
## Remote run: Unable to establish SSH connection
Automated ML uses the SSH protocol to communicate with remote DSVMs. This defaults to port 22. Possible causes for this error are:
1) The DSVM is not ready for SSH connections. When DSVM creation completes, the DSVM might still not be ready to acceept SSH connections. The sample notebooks have a one minute delay to allow for this.
2) Your Azure Subscription may restrict the IP address ranges that can access the DSVM on port 22. You can check this in the Azure Portal by selecting the Virtual Machine and then clicking Networking. The Virtual Machine name is the name that you provided in the notebook plus 10 alpha numeric characters to make the name unique. The Inbound Port Rules define what can access the VM on specific ports. Note that there is a priority priority order. So, a Deny entry with a low priority number will override a Allow entry with a higher priority number.
## Remote run: setup iteration fails
This is often an issue with the `get_data` method.
1) Check that the `get_data` method is valid by running it locally.
2) Make sure that `get_data` isn't referring to any local files. `get_data` is executed on the remote DSVM. So, it doesn't have direct access to local data files. Instead you can store the data files with DataStore. See [auto-ml-remote-execution-with-datastore.ipynb](remote-execution-with-datastore/auto-ml-remote-execution-with-datastore.ipynb)
3) You can get to the error log for the setup iteration by clicking the `Click here to see the run in Azure portal` link, click `Back to Experiment`, click on the highest run number and then click on Logs.
## Remote run: disk full
Automated ML creates files under /tmp/azureml_runs for each iteration that it runs. It creates a folder with the iteration id. For example: AutoML_9a038a18-77cc-48f1-80fb-65abdbc33abe_93. Under this, there is a azureml-logs folder, which contains logs. If you run too many iterations on the same DSVM, these files can fill the disk.
You can delete the files under /tmp/azureml_runs or just delete the VM and create a new one.
If your get_data downloads files, make sure the delete them or they can use disk space as well.
When using DataStore, it is good to specify an absolute path for the files so that they are downloaded just once. If you specify a relative path, it will download a file for each iteration.
## Remote run: Iterations fail and the log contains "MemoryError"
This can be caused by insufficient memory on the DSVM. Automated ML loads all training data into memory. So, the available memory should be more than the training data size.
If you are using a remote DSVM, memory is needed for each concurrent iteration. The max_concurrent_iterations setting specifies the maximum concurrent iterations. For example, if the training data size is 8Gb and max_concurrent_iterations is set to 10, the minimum memory required is at least 80Gb.
To resolve this issue, allocate a DSVM with more memory or reduce the value specified for max_concurrent_iterations.
## Remote run: Iterations show as "Not Responding" in the RunDetails widget.
This can be caused by too many concurrent iterations for a remote DSVM. Each concurrent iteration usually takes 100% of a core when it is running. Some iterations can use multiple cores. So, the max_concurrent_iterations setting should always be less than the number of cores of the DSVM.
To resolve this issue, try reducing the value specified for the max_concurrent_iterations setting.

30
how-to-use-azureml/automated-machine-learning/automl_env.yml Normal file
View File

@@ -0,0 +1,30 @@
name: azure_automl
channels:
- conda-forge
- pytorch
- main
dependencies:
# The python interpreter version.
# Currently Azure ML only supports 3.6.0 and later.
- pip==20.2.4
- python>=3.6,<3.9
- matplotlib==3.3.4
- py-xgboost==1.3.3
- pytorch::pytorch=1.4.0
- conda-forge::fbprophet==0.7.1
- cudatoolkit=10.1.243
- tqdm==4.63.1
- notebook
- pywin32==225
- PySocks==1.7.1
- conda-forge::pyqt==5.12.3
- pip:
# Required packages for AzureML execution, history, and data preparation.
- azureml-widgets~=1.40.0
- pytorch-transformers==1.0.0
- spacy==2.2.4
- pystan==2.19.1.1
- https://aka.ms/automl-resources/packages/en_core_web_sm-2.1.0.tar.gz
- -r https://automlsdkdataresources.blob.core.windows.net/validated-requirements/1.40.0/validated_win32_requirements.txt [--no-deps]
- arch==4.14

33
how-to-use-azureml/automated-machine-learning/automl_env_linux.yml Normal file
View File

@@ -0,0 +1,33 @@
name: azure_automl
channels:
- conda-forge
- pytorch
- main
dependencies:
# The python interpreter version.
# Currently Azure ML only supports 3.6.0 and later.
- pip==20.2.4
- python>=3.6,<3.9
- boto3==1.20.19
- botocore<=1.23.19
- matplotlib==3.3.4
- numpy==1.19.5
- cython==0.29.14
- urllib3==1.26.7
- scipy>=1.4.1,<=1.5.2
- scikit-learn==0.22.1
- py-xgboost<=1.3.3
- holidays==0.10.3
- conda-forge::fbprophet==0.7.1
- pytorch::pytorch=1.4.0
- cudatoolkit=10.1.243
- pip:
# Required packages for AzureML execution, history, and data preparation.
- azureml-widgets~=1.40.0
- pytorch-transformers==1.0.0
- spacy==2.2.4
- pystan==2.19.1.1
- https://aka.ms/automl-resources/packages/en_core_web_sm-2.1.0.tar.gz
- -r https://automlsdkdataresources.blob.core.windows.net/validated-requirements/1.40.0/validated_linux_requirements.txt [--no-deps]
- arch==4.14

34
how-to-use-azureml/automated-machine-learning/automl_env_mac.yml Normal file
View File

@@ -0,0 +1,34 @@
name: azure_automl
channels:
- conda-forge
- pytorch
- main
dependencies:
# The python interpreter version.
# Currently Azure ML only supports 3.6.0 and later.
- pip==20.2.4
- nomkl
- python>=3.6,<3.9
- boto3==1.20.19
- botocore<=1.23.19
- matplotlib==3.3.4
- numpy==1.19.5
- cython==0.29.14
- urllib3==1.26.7
- scipy>=1.4.1,<=1.5.2
- scikit-learn==0.22.1
- py-xgboost<=1.3.3
- holidays==0.10.3
- conda-forge::fbprophet==0.7.1
- pytorch::pytorch=1.4.0
- cudatoolkit=9.0
- pip:
# Required packages for AzureML execution, history, and data preparation.
- azureml-widgets~=1.40.0
- pytorch-transformers==1.0.0
- spacy==2.2.4
- pystan==2.19.1.1
- https://aka.ms/automl-resources/packages/en_core_web_sm-2.1.0.tar.gz
- -r https://automlsdkdataresources.blob.core.windows.net/validated-requirements/1.40.0/validated_darwin_requirements.txt [--no-deps]
- arch==4.14

78
how-to-use-azureml/automated-machine-learning/automl_setup.cmd Normal file
View File

@@ -0,0 +1,78 @@
@echo off
set conda_env_name=%1
set automl_env_file=%2
set options=%3
set PIP_NO_WARN_SCRIPT_LOCATION=0
IF "%conda_env_name%"=="" SET conda_env_name="azure_automl"
IF "%automl_env_file%"=="" SET automl_env_file="automl_env.yml"
SET check_conda_version_script="check_conda_version.py"
IF NOT EXIST %automl_env_file% GOTO YmlMissing
IF "%CONDA_EXE%"=="" GOTO CondaMissing
IF NOT EXIST %check_conda_version_script% GOTO VersionCheckMissing
python "%check_conda_version_script%"
IF errorlevel 1 GOTO ErrorExit:
SET replace_version_script="replace_latest_version.ps1"
IF EXIST %replace_version_script% (
powershell -file %replace_version_script% %automl_env_file%
)
call conda activate %conda_env_name% 2>nul:
if not errorlevel 1 (
echo Upgrading existing conda environment %conda_env_name%
call pip uninstall azureml-train-automl -y -q
call conda env update --name %conda_env_name% --file %automl_env_file%
if errorlevel 1 goto ErrorExit
) else (
call conda env create -f %automl_env_file% -n %conda_env_name%
)
call conda activate %conda_env_name% 2>nul:
if errorlevel 1 goto ErrorExit
call python -m ipykernel install --user --name %conda_env_name% --display-name "Python (%conda_env_name%)"
REM azureml.widgets is now installed as part of the pip install under the conda env.
REM Removing the old user install so that the notebooks will use the latest widget.
call jupyter nbextension uninstall --user --py azureml.widgets
echo.
echo.
echo ***************************************
echo * AutoML setup completed successfully *
echo ***************************************
IF NOT "%options%"=="nolaunch" (
echo.
echo Starting jupyter notebook - please run the configuration notebook
echo.
jupyter notebook --log-level=50 --notebook-dir='..\..'
)
goto End
:CondaMissing
echo Please run this script from an Anaconda Prompt window.
echo You can start an Anaconda Prompt window by
echo typing Anaconda Prompt on the Start menu.
echo If you don't see the Anaconda Prompt app, install Miniconda.
echo If you are running an older version of Miniconda or Anaconda,
echo you can upgrade using the command: conda update conda
goto End
:VersionCheckMissing
echo File %check_conda_version_script% not found.
goto End
:YmlMissing
echo File %automl_env_file% not found.
:ErrorExit
echo Install failed
:End

66
how-to-use-azureml/automated-machine-learning/automl_setup_linux.sh Normal file
View File

@@ -0,0 +1,66 @@
#!/bin/bash
CONDA_ENV_NAME=$1
AUTOML_ENV_FILE=$2
OPTIONS=$3
PIP_NO_WARN_SCRIPT_LOCATION=0
CHECK_CONDA_VERSION_SCRIPT="check_conda_version.py"
if [ "$CONDA_ENV_NAME" == "" ]
then
CONDA_ENV_NAME="azure_automl"
fi
if [ "$AUTOML_ENV_FILE" == "" ]
then
AUTOML_ENV_FILE="automl_env_linux.yml"
fi
if [ ! -f $AUTOML_ENV_FILE ]; then
echo "File $AUTOML_ENV_FILE not found"
exit 1
fi
if [ ! -f $CHECK_CONDA_VERSION_SCRIPT ]; then
echo "File $CHECK_CONDA_VERSION_SCRIPT not found"
exit 1
fi
python "$CHECK_CONDA_VERSION_SCRIPT"
if [ $? -ne 0 ]; then
exit 1
fi
sed -i 's/AZUREML-SDK-VERSION/latest/' $AUTOML_ENV_FILE
if source activate $CONDA_ENV_NAME 2> /dev/null
then
echo "Upgrading existing conda environment" $CONDA_ENV_NAME
pip uninstall azureml-train-automl -y -q
conda env update --name $CONDA_ENV_NAME --file $AUTOML_ENV_FILE &&
jupyter nbextension uninstall --user --py azureml.widgets
else
conda env create -f $AUTOML_ENV_FILE -n $CONDA_ENV_NAME &&
source activate $CONDA_ENV_NAME &&
python -m ipykernel install --user --name $CONDA_ENV_NAME --display-name "Python ($CONDA_ENV_NAME)" &&
jupyter nbextension uninstall --user --py azureml.widgets &&
echo "" &&
echo "" &&
echo "***************************************" &&
echo "* AutoML setup completed successfully *" &&
echo "***************************************" &&
if [ "$OPTIONS" != "nolaunch" ]
then
echo "" &&
echo "Starting jupyter notebook - please run the configuration notebook" &&
echo "" &&
jupyter notebook --log-level=50 --notebook-dir '../..'
fi
fi
if [ $? -gt 0 ]
then
echo "Installation failed"
fi

69
how-to-use-azureml/automated-machine-learning/automl_setup_mac.sh Normal file
View File

@@ -0,0 +1,69 @@
#!/bin/bash
CONDA_ENV_NAME=$1
AUTOML_ENV_FILE=$2
OPTIONS=$3
PIP_NO_WARN_SCRIPT_LOCATION=0
CHECK_CONDA_VERSION_SCRIPT="check_conda_version.py"
if [ "$CONDA_ENV_NAME" == "" ]
then
CONDA_ENV_NAME="azure_automl"
fi
if [ "$AUTOML_ENV_FILE" == "" ]
then
AUTOML_ENV_FILE="automl_env_mac.yml"
fi
if [ ! -f $AUTOML_ENV_FILE ]; then
echo "File $AUTOML_ENV_FILE not found"
exit 1
fi
if [ ! -f $CHECK_CONDA_VERSION_SCRIPT ]; then
echo "File $CHECK_CONDA_VERSION_SCRIPT not found"
exit 1
fi
python "$CHECK_CONDA_VERSION_SCRIPT"
if [ $? -ne 0 ]; then
exit 1
fi
sed -i '' 's/AZUREML-SDK-VERSION/latest/' $AUTOML_ENV_FILE
brew install libomp
if source activate $CONDA_ENV_NAME 2> /dev/null
then
echo "Upgrading existing conda environment" $CONDA_ENV_NAME
pip uninstall azureml-train-automl -y -q
conda env update --name $CONDA_ENV_NAME --file $AUTOML_ENV_FILE &&
jupyter nbextension uninstall --user --py azureml.widgets
else
conda env create -f $AUTOML_ENV_FILE -n $CONDA_ENV_NAME &&
source activate $CONDA_ENV_NAME &&
conda install lightgbm -c conda-forge -y &&
python -m ipykernel install --user --name $CONDA_ENV_NAME --display-name "Python ($CONDA_ENV_NAME)" &&
jupyter nbextension uninstall --user --py azureml.widgets &&
echo "" &&
echo "" &&
echo "***************************************" &&
echo "* AutoML setup completed successfully *" &&
echo "***************************************" &&
if [ "$OPTIONS" != "nolaunch" ]
then
echo "" &&
echo "Starting jupyter notebook - please run the configuration notebook" &&
echo "" &&
jupyter notebook --log-level=50 --notebook-dir '../..'
fi
fi
if [ $? -gt 0 ]
then
echo "Installation failed"
fi

26
how-to-use-azureml/automated-machine-learning/check_conda_version.py Normal file
View File

@@ -0,0 +1,26 @@
from distutils.version import LooseVersion
import platform
try:
import conda
except Exception:
print('Failed to import conda.')
print('This setup is usually run from the base conda environment.')
print('You can activate the base environment using the command "conda activate base"')
exit(1)
architecture = platform.architecture()[0]
if architecture != "64bit":
print('This setup requires 64bit Anaconda or Miniconda. Found: ' + architecture)
exit(1)
minimumVersion = "4.7.8"
versionInvalid = (LooseVersion(conda.__version__) < LooseVersion(minimumVersion))
if versionInvalid:
print('Setup requires conda version ' + minimumVersion + ' or higher.')
print('You can use the command "conda update conda" to upgrade conda.')
exit(versionInvalid)

1091
how-to-use-azureml/automated-machine-learning/classification-bank-marketing-all-features/auto-ml-classification-bank-marketing-all-features.ipynb Normal file
View File

File diff suppressed because it is too large Load Diff

4
how-to-use-azureml/automated-machine-learning/classification-bank-marketing-all-features/auto-ml-classification-bank-marketing-all-features.yml Normal file
View File

@@ -0,0 +1,4 @@
name: auto-ml-classification-bank-marketing-all-features
dependencies:
- pip:
- azureml-sdk

487
how-to-use-azureml/automated-machine-learning/classification-credit-card-fraud/auto-ml-classification-credit-card-fraud.ipynb Normal file
View File

@@ -0,0 +1,487 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Automated Machine Learning\n",
"_**Classification of credit card fraudulent transactions on remote compute **_\n",
"\n",
"## Contents\n",
"1. [Introduction](#Introduction)\n",
"1. [Setup](#Setup)\n",
"1. [Train](#Train)\n",
"1. [Results](#Results)\n",
"1. [Test](#Test)\n",
"1. [Acknowledgements](#Acknowledgements)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Introduction\n",
"\n",
"In this example we use the associated credit card dataset to showcase how you can use AutoML for a simple classification problem. The goal is to predict if a credit card transaction is considered a fraudulent charge.\n",
"\n",
"This notebook is using remote compute to train the model.\n",
"\n",
"If you are using an Azure Machine Learning Compute Instance, you are all set. Otherwise, go through the [configuration](../../../configuration.ipynb) notebook first if you haven't already to establish your connection to the AzureML Workspace. \n",
"\n",
"In this notebook you will learn how to:\n",
"1. Create an experiment using an existing workspace.\n",
"2. Configure AutoML using `AutoMLConfig`.\n",
"3. Train the model using remote compute.\n",
"4. Explore the results.\n",
"5. Test the fitted model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Setup\n",
"\n",
"As part of the setup you have already created an Azure ML `Workspace` object. For Automated ML you will need to create an `Experiment` object, which is a named object in a `Workspace` used to run experiments."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import logging\n",
"\n",
"from matplotlib import pyplot as plt\n",
"import pandas as pd\n",
"import os\n",
"\n",
"import azureml.core\n",
"from azureml.core.experiment import Experiment\n",
"from azureml.core.workspace import Workspace\n",
"from azureml.core.dataset import Dataset\n",
"from azureml.train.automl import AutoMLConfig"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This sample notebook may use features that are not available in previous versions of the Azure ML SDK."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"ws = Workspace.from_config()\n",
"\n",
"# choose a name for experiment\n",
"experiment_name = \"automl-classification-ccard-remote\"\n",
"\n",
"experiment = Experiment(ws, experiment_name)\n",
"\n",
"output = {}\n",
"output[\"Subscription ID\"] = ws.subscription_id\n",
"output[\"Workspace\"] = ws.name\n",
"output[\"Resource Group\"] = ws.resource_group\n",
"output[\"Location\"] = ws.location\n",
"output[\"Experiment Name\"] = experiment.name\n",
"pd.set_option(\"display.max_colwidth\", None)\n",
"outputDf = pd.DataFrame(data=output, index=[\"\"])\n",
"outputDf.T"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Create or Attach existing AmlCompute\n",
"A compute target is required to execute the Automated ML run. In this tutorial, you create AmlCompute as your training compute resource.\n",
"\n",
"> Note that if you have an AzureML Data Scientist role, you will not have permission to create compute resources. Talk to your workspace or IT admin to create the compute targets described in this section, if they do not already exist.\n",
"\n",
"#### Creation of AmlCompute takes approximately 5 minutes. \n",
"If the AmlCompute with that name is already in your workspace this code will skip the creation process.\n",
"As with other Azure services, there are limits on certain resources (e.g. AmlCompute) associated with the Azure Machine Learning service. Please read [this article](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-manage-quotas) on the default limits and how to request more quota."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core.compute import ComputeTarget, AmlCompute\n",
"from azureml.core.compute_target import ComputeTargetException\n",
"\n",
"# Choose a name for your CPU cluster\n",
"cpu_cluster_name = \"cpu-cluster-1\"\n",
"\n",
"# Verify that cluster does not exist already\n",
"try:\n",
" compute_target = ComputeTarget(workspace=ws, name=cpu_cluster_name)\n",
" print(\"Found existing cluster, use it.\")\n",
"except ComputeTargetException:\n",
" compute_config = AmlCompute.provisioning_configuration(\n",
" vm_size=\"STANDARD_DS12_V2\", max_nodes=6\n",
" )\n",
" compute_target = ComputeTarget.create(ws, cpu_cluster_name, compute_config)\n",
"compute_target.wait_for_completion(show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Data"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Load Data\n",
"\n",
"Load the credit card dataset from a csv file containing both training features and labels. The features are inputs to the model, while the training labels represent the expected output of the model. Next, we'll split the data using random_split and extract the training data for the model."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"name": "load-data"
},
"outputs": [],
"source": [
"data = \"https://automlsamplenotebookdata.blob.core.windows.net/automl-sample-notebook-data/creditcard.csv\"\n",
"dataset = Dataset.Tabular.from_delimited_files(data)\n",
"training_data, validation_data = dataset.random_split(percentage=0.8, seed=223)\n",
"label_column_name = \"Class\""
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Train\n",
"\n",
"Instantiate a AutoMLConfig object. This defines the settings and data used to run the experiment.\n",
"\n",
"|Property|Description|\n",
"|-|-|\n",
"|**task**|classification or regression|\n",
"|**primary_metric**|This is the metric that you want to optimize. Classification supports the following primary metrics: <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>average_precision_score_weighted</i><br><i>norm_macro_recall</i><br><i>precision_score_weighted</i>|\n",
"|**enable_early_stopping**|Stop the run if the metric score is not showing improvement.|\n",
"|**n_cross_validations**|Number of cross validation splits.|\n",
"|**training_data**|Input dataset, containing both features and label column.|\n",
"|**label_column_name**|The name of the label column.|\n",
"\n",
"**_You can find more information about primary metrics_** [here](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-configure-auto-train#primary-metric)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"name": "automl-config"
},
"outputs": [],
"source": [
"automl_settings = {\n",
" \"n_cross_validations\": 3,\n",
" \"primary_metric\": \"average_precision_score_weighted\",\n",
" \"enable_early_stopping\": True,\n",
" \"max_concurrent_iterations\": 2, # This is a limit for testing purpose, please increase it as per cluster size\n",
" \"experiment_timeout_hours\": 0.25, # This is a time limit for testing purposes, remove it for real use cases, this will drastically limit ablity to find the best model possible\n",
" \"verbosity\": logging.INFO,\n",
"}\n",
"\n",
"automl_config = AutoMLConfig(\n",
" task=\"classification\",\n",
" debug_log=\"automl_errors.log\",\n",
" compute_target=compute_target,\n",
" training_data=training_data,\n",
" label_column_name=label_column_name,\n",
" **automl_settings,\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Call the `submit` method on the experiment object and pass the run configuration. Depending on the data and the number of iterations this can run for a while. Validation errors and current status will be shown when setting `show_output=True` and the execution will be synchronous."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"remote_run = experiment.submit(automl_config, show_output=False)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# If you need to retrieve a run that already started, use the following code\n",
"# from azureml.train.automl.run import AutoMLRun\n",
"# remote_run = AutoMLRun(experiment = experiment, run_id = '<replace with your run id>')"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Results"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Widget for Monitoring Runs\n",
"\n",
"The widget will first report a \"loading\" status while running the first iteration. After completing the first iteration, an auto-updating graph and table will be shown. The widget will refresh once per minute, so you should see the graph update as child runs complete.\n",
"\n",
"**Note:** The widget displays a link at the bottom. Use this link to open a web interface to explore the individual run details"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"tags": [
"widget-rundetails-sample"
]
},
"outputs": [],
"source": [
"from azureml.widgets import RunDetails\n",
"\n",
"RunDetails(remote_run).show()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"remote_run.wait_for_completion(show_output=False)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Explain model\n",
"\n",
"Automated ML models can be explained and visualized using the SDK Explainability library. "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Analyze results\n",
"\n",
"### Retrieve the Best Model\n",
"\n",
"Below we select the best pipeline from our iterations. The `get_output` method returns the best run and the fitted model. Overloads on `get_output` allow you to retrieve the best run and fitted model for *any* logged metric or for a particular *iteration*."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"best_run, fitted_model = remote_run.get_output()\n",
"fitted_model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Print the properties of the model\n",
"The fitted_model is a python object and you can read the different properties of the object.\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Test the fitted model\n",
"\n",
"Now that the model is trained, split the data in the same way the data was split for training (The difference here is the data is being split locally) and then run the test data through the trained model to get the predicted values."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# convert the test data to dataframe\n",
"X_test_df = validation_data.drop_columns(\n",
" columns=[label_column_name]\n",
").to_pandas_dataframe()\n",
"y_test_df = validation_data.keep_columns(\n",
" columns=[label_column_name], validate=True\n",
").to_pandas_dataframe()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# call the predict functions on the model\n",
"y_pred = fitted_model.predict(X_test_df)\n",
"y_pred"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Calculate metrics for the prediction\n",
"\n",
"Now visualize the data on a scatter plot to show what our truth (actual) values are compared to the predicted values \n",
"from the trained model that was returned."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from sklearn.metrics import confusion_matrix\n",
"import numpy as np\n",
"import itertools\n",
"\n",
"cf = confusion_matrix(y_test_df.values, y_pred)\n",
"plt.imshow(cf, cmap=plt.cm.Blues, interpolation=\"nearest\")\n",
"plt.colorbar()\n",
"plt.title(\"Confusion Matrix\")\n",
"plt.xlabel(\"Predicted\")\n",
"plt.ylabel(\"Actual\")\n",
"class_labels = [\"False\", \"True\"]\n",
"tick_marks = np.arange(len(class_labels))\n",
"plt.xticks(tick_marks, class_labels)\n",
"plt.yticks([-0.5, 0, 1, 1.5], [\"\", \"False\", \"True\", \"\"])\n",
"# plotting text value inside cells\n",
"thresh = cf.max() / 2.0\n",
"for i, j in itertools.product(range(cf.shape[0]), range(cf.shape[1])):\n",
" plt.text(\n",
" j,\n",
" i,\n",
" format(cf[i, j], \"d\"),\n",
" horizontalalignment=\"center\",\n",
" color=\"white\" if cf[i, j] > thresh else \"black\",\n",
" )\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Acknowledgements"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This Credit Card fraud Detection dataset is made available under the Open Database License: http://opendatacommons.org/licenses/odbl/1.0/. Any rights in individual contents of the database are licensed under the Database Contents License: http://opendatacommons.org/licenses/dbcl/1.0/ and is available at: https://www.kaggle.com/mlg-ulb/creditcardfraud\n",
"\n",
"The dataset has been collected and analysed during a research collaboration of Worldline and the Machine Learning Group (http://mlg.ulb.ac.be) of ULB (Universit\u00c3\u00a9 Libre de Bruxelles) on big data mining and fraud detection.\n",
"More details on current and past projects on related topics are available on https://www.researchgate.net/project/Fraud-detection-5 and the page of the DefeatFraud project\n",
"\n",
"Please cite the following works:\n",
"\n",
"Andrea Dal Pozzolo, Olivier Caelen, Reid A. Johnson and Gianluca Bontempi. Calibrating Probability with Undersampling for Unbalanced Classification. In Symposium on Computational Intelligence and Data Mining (CIDM), IEEE, 2015\n",
"\n",
"Dal Pozzolo, Andrea; Caelen, Olivier; Le Borgne, Yann-Ael; Waterschoot, Serge; Bontempi, Gianluca. Learned lessons in credit card fraud detection from a practitioner perspective, Expert systems with applications,41,10,4915-4928,2014, Pergamon\n",
"\n",
"Dal Pozzolo, Andrea; Boracchi, Giacomo; Caelen, Olivier; Alippi, Cesare; Bontempi, Gianluca. Credit card fraud detection: a realistic modeling and a novel learning strategy, IEEE transactions on neural networks and learning systems,29,8,3784-3797,2018,IEEE\n",
"\n",
"Dal Pozzolo, Andrea Adaptive Machine learning for credit card fraud detection ULB MLG PhD thesis (supervised by G. Bontempi)\n",
"\n",
"Carcillo, Fabrizio; Dal Pozzolo, Andrea; Le Borgne, Yann-A\u00c3\u00abl; Caelen, Olivier; Mazzer, Yannis; Bontempi, Gianluca. Scarff: a scalable framework for streaming credit card fraud detection with Spark, Information fusion,41, 182-194,2018,Elsevier\n",
"\n",
"Carcillo, Fabrizio; Le Borgne, Yann-A\u00c3\u00abl; Caelen, Olivier; Bontempi, Gianluca. Streaming active learning strategies for real-life credit card fraud detection: assessment and visualization, International Journal of Data Science and Analytics, 5,4,285-300,2018,Springer International Publishing\n",
"\n",
"Bertrand Lebichot, Yann-A\u00c3\u00abl Le Borgne, Liyun He, Frederic Obl\u00c3\u00a9, Gianluca Bontempi Deep-Learning Domain Adaptation Techniques for Credit Cards Fraud Detection, INNSBDDL 2019: Recent Advances in Big Data and Deep Learning, pp 78-88, 2019\n",
"\n",
"Fabrizio Carcillo, Yann-A\u00c3\u00abl Le Borgne, Olivier Caelen, Frederic Obl\u00c3\u00a9, Gianluca Bontempi Combining Unsupervised and Supervised Learning in Credit Card Fraud Detection Information Sciences, 2019"
]
}
],
"metadata": {
"authors": [
{
"name": "ratanase"
}
],
"category": "tutorial",
"compute": [
"AML Compute"
],
"datasets": [
"Creditcard"
],
"deployment": [
"None"
],
"exclude_from_index": false,
"file_extension": ".py",
"framework": [
"None"
],
"friendly_name": "Classification of credit card fraudulent transactions using Automated ML",
"index_order": 5,
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.7"
},
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"tags": [
"remote_run",
"AutomatedML"
],
"task": "Classification",
"version": "3.6.7"
},
"nbformat": 4,
"nbformat_minor": 2
}

4
how-to-use-azureml/automated-machine-learning/classification-credit-card-fraud/auto-ml-classification-credit-card-fraud.yml Normal file
View File

@@ -0,0 +1,4 @@
name: auto-ml-classification-credit-card-fraud
dependencies:
- pip:
- azureml-sdk

592
how-to-use-azureml/automated-machine-learning/classification-text-dnn/auto-ml-classification-text-dnn.ipynb Normal file
View File

@@ -0,0 +1,592 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Automated Machine Learning\n",
"_**Text Classification Using Deep Learning**_\n",
"\n",
"## Contents\n",
"1. [Introduction](#Introduction)\n",
"1. [Setup](#Setup)\n",
"1. [Data](#Data)\n",
"1. [Train](#Train)\n",
"1. [Evaluate](#Evaluate)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Introduction\n",
"This notebook demonstrates classification with text data using deep learning in AutoML.\n",
"\n",
"AutoML highlights here include using deep neural networks (DNNs) to create embedded features from text data. Depending on the compute cluster the user provides, AutoML tried out Bidirectional Encoder Representations from Transformers (BERT) when a GPU compute is used, and Bidirectional Long-Short Term neural network (BiLSTM) when a CPU compute is used, thereby optimizing the choice of DNN for the uesr's setup.\n",
"\n",
"Make sure you have executed the [configuration](../../../configuration.ipynb) before running this notebook.\n",
"\n",
"Notebook synopsis:\n",
"\n",
"1. Creating an Experiment in an existing Workspace\n",
"2. Configuration and remote run of AutoML for a text dataset (20 Newsgroups dataset from scikit-learn) for classification\n",
"3. Registering the best model for future use\n",
"4. Evaluating the final model on a test set"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Setup"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import json\n",
"import logging\n",
"import os\n",
"import shutil\n",
"\n",
"import pandas as pd\n",
"\n",
"import azureml.core\n",
"from azureml.core.experiment import Experiment\n",
"from azureml.core.workspace import Workspace\n",
"from azureml.core.dataset import Dataset\n",
"from azureml.core.compute import AmlCompute\n",
"from azureml.core.compute import ComputeTarget\n",
"from azureml.core.run import Run\n",
"from azureml.widgets import RunDetails\n",
"from azureml.core.model import Model\n",
"from helper import run_inference, get_result_df\n",
"from azureml.train.automl import AutoMLConfig\n",
"from sklearn.datasets import fetch_20newsgroups"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This sample notebook may use features that are not available in previous versions of the Azure ML SDK."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"As part of the setup you have already created a <b>Workspace</b>. To run AutoML, you also need to create an <b>Experiment</b>. An Experiment corresponds to a prediction problem you are trying to solve, while a Run corresponds to a specific approach to the problem."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"ws = Workspace.from_config()\n",
"\n",
"# Choose an experiment name.\n",
"experiment_name = \"automl-classification-text-dnn\"\n",
"\n",
"experiment = Experiment(ws, experiment_name)\n",
"\n",
"output = {}\n",
"output[\"Subscription ID\"] = ws.subscription_id\n",
"output[\"Workspace Name\"] = ws.name\n",
"output[\"Resource Group\"] = ws.resource_group\n",
"output[\"Location\"] = ws.location\n",
"output[\"Experiment Name\"] = experiment.name\n",
"pd.set_option(\"display.max_colwidth\", None)\n",
"outputDf = pd.DataFrame(data=output, index=[\"\"])\n",
"outputDf.T"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Set up a compute cluster\n",
"This section uses a user-provided compute cluster (named \"dnntext-cluster\" in this example). If a cluster with this name does not exist in the user's workspace, the below code will create a new cluster. You can choose the parameters of the cluster as mentioned in the comments.\n",
"\n",
"> Note that if you have an AzureML Data Scientist role, you will not have permission to create compute resources. Talk to your workspace or IT admin to create the compute targets described in this section, if they do not already exist.\n",
"\n",
"Whether you provide/select a CPU or GPU cluster, AutoML will choose the appropriate DNN for that setup - BiLSTM or BERT text featurizer will be included in the candidate featurizers on CPU and GPU respectively. If your goal is to obtain the most accurate model, we recommend you use GPU clusters since BERT featurizers usually outperform BiLSTM featurizers."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core.compute import ComputeTarget, AmlCompute\n",
"from azureml.core.compute_target import ComputeTargetException\n",
"\n",
"num_nodes = 2\n",
"\n",
"# Choose a name for your cluster.\n",
"amlcompute_cluster_name = \"dnntext-cluster\"\n",
"\n",
"# Verify that cluster does not exist already\n",
"try:\n",
" compute_target = ComputeTarget(workspace=ws, name=amlcompute_cluster_name)\n",
" print(\"Found existing cluster, use it.\")\n",
"except ComputeTargetException:\n",
" compute_config = AmlCompute.provisioning_configuration(\n",
" vm_size=\"STANDARD_NC6\", # CPU for BiLSTM, such as \"STANDARD_D2_V2\"\n",
" # To use BERT (this is recommended for best performance), select a GPU such as \"STANDARD_NC6\"\n",
" # or similar GPU option\n",
" # available in your workspace\n",
" idle_seconds_before_scaledown=60,\n",
" max_nodes=num_nodes,\n",
" )\n",
" compute_target = ComputeTarget.create(ws, amlcompute_cluster_name, compute_config)\n",
"\n",
"compute_target.wait_for_completion(show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Get data\n",
"For this notebook we will use 20 Newsgroups data from scikit-learn. We filter the data to contain four classes and take a sample as training data. Please note that for accuracy improvement, more data is needed. For this notebook we provide a small-data example so that you can use this template to use with your larger sized data."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"data_dir = \"text-dnn-data\" # Local directory to store data\n",
"blobstore_datadir = data_dir # Blob store directory to store data in\n",
"target_column_name = \"y\"\n",
"feature_column_name = \"X\"\n",
"\n",
"\n",
"def get_20newsgroups_data():\n",
" \"\"\"Fetches 20 Newsgroups data from scikit-learn\n",
" Returns them in form of pandas dataframes\n",
" \"\"\"\n",
" remove = (\"headers\", \"footers\", \"quotes\")\n",
" categories = [\n",
" \"rec.sport.baseball\",\n",
" \"rec.sport.hockey\",\n",
" \"comp.graphics\",\n",
" \"sci.space\",\n",
" ]\n",
"\n",
" data = fetch_20newsgroups(\n",
" subset=\"train\",\n",
" categories=categories,\n",
" shuffle=True,\n",
" random_state=42,\n",
" remove=remove,\n",
" )\n",
" data = pd.DataFrame(\n",
" {feature_column_name: data.data, target_column_name: data.target}\n",
" )\n",
"\n",
" data_train = data[:200]\n",
" data_test = data[200:300]\n",
"\n",
" data_train = remove_blanks_20news(\n",
" data_train, feature_column_name, target_column_name\n",
" )\n",
" data_test = remove_blanks_20news(data_test, feature_column_name, target_column_name)\n",
"\n",
" return data_train, data_test\n",
"\n",
"\n",
"def remove_blanks_20news(data, feature_column_name, target_column_name):\n",
"\n",
" data[feature_column_name] = (\n",
" data[feature_column_name]\n",
" .replace(r\"\\n\", \" \", regex=True)\n",
" .apply(lambda x: x.strip())\n",
" )\n",
" data = data[data[feature_column_name] != \"\"]\n",
"\n",
" return data"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Fetch data and upload to datastore for use in training"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"data_train, data_test = get_20newsgroups_data()\n",
"\n",
"if not os.path.isdir(data_dir):\n",
" os.mkdir(data_dir)\n",
"\n",
"train_data_fname = data_dir + \"/train_data.csv\"\n",
"test_data_fname = data_dir + \"/test_data.csv\"\n",
"\n",
"data_train.to_csv(train_data_fname, index=False)\n",
"data_test.to_csv(test_data_fname, index=False)\n",
"\n",
"datastore = ws.get_default_datastore()\n",
"datastore.upload(src_dir=data_dir, target_path=blobstore_datadir, overwrite=True)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"train_dataset = Dataset.Tabular.from_delimited_files(\n",
" path=[(datastore, blobstore_datadir + \"/train_data.csv\")]\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Prepare AutoML run"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This notebook uses the blocked_models parameter to exclude some models that can take a longer time to train on some text datasets. You can choose to remove models from the blocked_models list but you may need to increase the experiment_timeout_hours parameter value to get results."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"automl_settings = {\n",
" \"experiment_timeout_minutes\": 30,\n",
" \"primary_metric\": \"accuracy\",\n",
" \"max_concurrent_iterations\": num_nodes,\n",
" \"max_cores_per_iteration\": -1,\n",
" \"enable_dnn\": True,\n",
" \"enable_early_stopping\": True,\n",
" \"validation_size\": 0.3,\n",
" \"verbosity\": logging.INFO,\n",
" \"enable_voting_ensemble\": False,\n",
" \"enable_stack_ensemble\": False,\n",
"}\n",
"\n",
"automl_config = AutoMLConfig(\n",
" task=\"classification\",\n",
" debug_log=\"automl_errors.log\",\n",
" compute_target=compute_target,\n",
" training_data=train_dataset,\n",
" label_column_name=target_column_name,\n",
" blocked_models=[\"LightGBM\", \"XGBoostClassifier\"],\n",
" **automl_settings,\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Submit AutoML Run"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"automl_run = experiment.submit(automl_config, show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Displaying the run objects gives you links to the visual tools in the Azure Portal. Go try them!"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Retrieve the Best Model\n",
"Below we select the best model pipeline from our iterations, use it to test on test data on the same compute cluster."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"For local inferencing, you can load the model locally via. the method `remote_run.get_output()`. For more information on the arguments expected by this method, you can run `remote_run.get_output??`.\n",
"Note that when the model contains BERT, this step will require pytorch and pytorch-transformers installed in your local environment. The exact versions of these packages can be found in the **automl_env.yml** file located in the local copy of your azureml-examples folder here: \"azureml-examples/python-sdk/tutorials/automl-with-azureml\""
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Retrieve the best Run object\n",
"best_run = automl_run.get_best_child()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"You can now see what text transformations are used to convert text data to features for this dataset, including deep learning transformations based on BiLSTM or Transformer (BERT is one implementation of a Transformer) models."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Download the featurization summary JSON file locally\n",
"best_run.download_file(\n",
" \"outputs/featurization_summary.json\", \"featurization_summary.json\"\n",
")\n",
"\n",
"# Render the JSON as a pandas DataFrame\n",
"with open(\"featurization_summary.json\", \"r\") as f:\n",
" records = json.load(f)\n",
"\n",
"featurization_summary = pd.DataFrame.from_records(records)\n",
"featurization_summary[\"Transformations\"].tolist()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Registering the best model\n",
"We now register the best fitted model from the AutoML Run for use in future deployments. "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Get results stats, extract the best model from AutoML run, download and register the resultant best model"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"summary_df = get_result_df(automl_run)\n",
"best_dnn_run_id = summary_df[\"run_id\"].iloc[0]\n",
"best_dnn_run = Run(experiment, best_dnn_run_id)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"model_dir = \"Model\" # Local folder where the model will be stored temporarily\n",
"if not os.path.isdir(model_dir):\n",
" os.mkdir(model_dir)\n",
"\n",
"best_dnn_run.download_file(\"outputs/model.pkl\", model_dir + \"/model.pkl\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Register the model in your Azure Machine Learning Workspace. If you previously registered a model, please make sure to delete it so as to replace it with this new model."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Register the model\n",
"model_name = \"textDNN-20News\"\n",
"model = Model.register(\n",
" model_path=model_dir + \"/model.pkl\", model_name=model_name, tags=None, workspace=ws\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Evaluate on Test Data"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We now use the best fitted model from the AutoML Run to make predictions on the test set. \n",
"\n",
"Test set schema should match that of the training set."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"test_dataset = Dataset.Tabular.from_delimited_files(\n",
" path=[(datastore, blobstore_datadir + \"/test_data.csv\")]\n",
")\n",
"\n",
"# preview the first 3 rows of the dataset\n",
"test_dataset.take(3).to_pandas_dataframe()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"test_experiment = Experiment(ws, experiment_name + \"_test\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"script_folder = os.path.join(os.getcwd(), \"inference\")\n",
"os.makedirs(script_folder, exist_ok=True)\n",
"shutil.copy(\"infer.py\", script_folder)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"test_run = run_inference(\n",
" test_experiment,\n",
" compute_target,\n",
" script_folder,\n",
" best_dnn_run,\n",
" test_dataset,\n",
" target_column_name,\n",
" model_name,\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Display computed metrics"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"test_run"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"RunDetails(test_run).show()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"test_run.wait_for_completion()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"pd.Series(test_run.get_metrics())"
]
}
],
"metadata": {
"authors": [
{
"name": "anshirga"
}
],
"compute": [
"AML Compute"
],
"datasets": [
"None"
],
"deployment": [
"None"
],
"exclude_from_index": false,
"framework": [
"None"
],
"friendly_name": "DNN Text Featurization",
"index_order": 2,
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.7"
},
"tags": [
"None"
],
"task": "Text featurization using DNNs for classification"
},
"nbformat": 4,
"nbformat_minor": 2
}

4
how-to-use-azureml/automated-machine-learning/classification-text-dnn/auto-ml-classification-text-dnn.yml Normal file
View File

@@ -0,0 +1,4 @@
name: auto-ml-classification-text-dnn
dependencies:
- pip:
- azureml-sdk

68
how-to-use-azureml/automated-machine-learning/classification-text-dnn/helper.py Normal file
View File

@@ -0,0 +1,68 @@
import pandas as pd
from azureml.core import Environment
from azureml.train.estimator import Estimator
from azureml.core.run import Run
def run_inference(
test_experiment,
compute_target,
script_folder,
train_run,
test_dataset,
target_column_name,
model_name,
):
inference_env = train_run.get_environment()
est = Estimator(
source_directory=script_folder,
entry_script="infer.py",
script_params={
"--target_column_name": target_column_name,
"--model_name": model_name,
},
inputs=[test_dataset.as_named_input("test_data")],
compute_target=compute_target,
environment_definition=inference_env,
)
run = test_experiment.submit(
est,
tags={
"training_run_id": train_run.id,
"run_algorithm": train_run.properties["run_algorithm"],
"valid_score": train_run.properties["score"],
"primary_metric": train_run.properties["primary_metric"],
},
)
run.log("run_algorithm", run.tags["run_algorithm"])
return run
def get_result_df(remote_run):
children = list(remote_run.get_children(recursive=True))
summary_df = pd.DataFrame(
index=["run_id", "run_algorithm", "primary_metric", "Score"]
)
goal_minimize = False
for run in children:
if "run_algorithm" in run.properties and "score" in run.properties:
summary_df[run.id] = [
run.id,
run.properties["run_algorithm"],
run.properties["primary_metric"],
float(run.properties["score"]),
]
if "goal" in run.properties:
goal_minimize = run.properties["goal"].split("_")[-1] == "min"
summary_df = summary_df.T.sort_values(
"Score", ascending=goal_minimize
).drop_duplicates(["run_algorithm"])
summary_df = summary_df.set_index("run_algorithm")
return summary_df

68
how-to-use-azureml/automated-machine-learning/classification-text-dnn/infer.py Normal file
View File

@@ -0,0 +1,68 @@
import argparse
import pandas as pd
import numpy as np
from sklearn.externals import joblib
from azureml.automl.runtime.shared.score import scoring, constants
from azureml.core import Run
from azureml.core.model import Model
parser = argparse.ArgumentParser()
parser.add_argument(
"--target_column_name",
type=str,
dest="target_column_name",
help="Target Column Name",
)
parser.add_argument(
"--model_name", type=str, dest="model_name", help="Name of registered model"
)
args = parser.parse_args()
target_column_name = args.target_column_name
model_name = args.model_name
print("args passed are: ")
print("Target column name: ", target_column_name)
print("Name of registered model: ", model_name)
model_path = Model.get_model_path(model_name)
# deserialize the model file back into a sklearn model
model = joblib.load(model_path)
run = Run.get_context()
# get input dataset by name
test_dataset = run.input_datasets["test_data"]
X_test_df = test_dataset.drop_columns(
columns=[target_column_name]
).to_pandas_dataframe()
y_test_df = (
test_dataset.with_timestamp_columns(None)
.keep_columns(columns=[target_column_name])
.to_pandas_dataframe()
)
predicted = model.predict_proba(X_test_df)
if isinstance(predicted, pd.DataFrame):
predicted = predicted.values
# Use the AutoML scoring module
train_labels = model.classes_
class_labels = np.unique(
np.concatenate((y_test_df.values, np.reshape(train_labels, (-1, 1))))
)
classification_metrics = list(constants.CLASSIFICATION_SCALAR_SET)
scores = scoring.score_classification(
y_test_df.values, predicted, classification_metrics, class_labels, train_labels
)
print("scores:")
print(scores)
for key, value in scores.items():
run.log(key, value)

585
how-to-use-azureml/automated-machine-learning/continuous-retraining/auto-ml-continuous-retraining.ipynb Normal file
View File

@@ -0,0 +1,585 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Automated Machine Learning \n",
"**Continuous retraining using Pipelines and Time-Series TabularDataset**\n",
"## Contents\n",
"1. [Introduction](#Introduction)\n",
"2. [Setup](#Setup)\n",
"3. [Compute](#Compute)\n",
"4. [Run Configuration](#Run-Configuration)\n",
"5. [Data Ingestion Pipeline](#Data-Ingestion-Pipeline)\n",
"6. [Training Pipeline](#Training-Pipeline)\n",
"7. [Publish Retraining Pipeline and Schedule](#Publish-Retraining-Pipeline-and-Schedule)\n",
"8. [Test Retraining](#Test-Retraining)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Introduction\n",
"In this example we use AutoML and Pipelines to enable contious retraining of a model based on updates to the training dataset. We will create two pipelines, the first one to demonstrate a training dataset that gets updated over time. We leverage time-series capabilities of `TabularDataset` to achieve this. The second pipeline utilizes pipeline `Schedule` to trigger continuous retraining. \n",
"Make sure you have executed the [configuration notebook](../../../configuration.ipynb) before running this notebook.\n",
"In this notebook you will learn how to:\n",
"* Create an Experiment in an existing Workspace.\n",
"* Configure AutoML using AutoMLConfig.\n",
"* Create data ingestion pipeline to update a time-series based TabularDataset\n",
"* Create training pipeline to prepare data, run AutoML, register the model and setup pipeline triggers.\n",
"\n",
"## Setup\n",
"As part of the setup you have already created an Azure ML `Workspace` object. For AutoML you will need to create an `Experiment` object, which is a named object in a `Workspace` used to run experiments."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import logging\n",
"\n",
"from matplotlib import pyplot as plt\n",
"import numpy as np\n",
"import pandas as pd\n",
"from sklearn import datasets\n",
"\n",
"import azureml.core\n",
"from azureml.core.experiment import Experiment\n",
"from azureml.core.workspace import Workspace\n",
"from azureml.train.automl import AutoMLConfig"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This sample notebook may use features that are not available in previous versions of the Azure ML SDK."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Accessing the Azure ML workspace requires authentication with Azure.\n",
"\n",
"The default authentication is interactive authentication using the default tenant. Executing the ws = Workspace.from_config() line in the cell below will prompt for authentication the first time that it is run.\n",
"\n",
"If you have multiple Azure tenants, you can specify the tenant by replacing the ws = Workspace.from_config() line in the cell below with the following:\n",
"```\n",
"from azureml.core.authentication import InteractiveLoginAuthentication\n",
"auth = InteractiveLoginAuthentication(tenant_id = 'mytenantid')\n",
"ws = Workspace.from_config(auth = auth)\n",
"```\n",
"If you need to run in an environment where interactive login is not possible, you can use Service Principal authentication by replacing the ws = Workspace.from_config() line in the cell below with the following:\n",
"```\n",
"from azureml.core.authentication import ServicePrincipalAuthentication\n",
"auth = auth = ServicePrincipalAuthentication('mytenantid', 'myappid', 'mypassword')\n",
"ws = Workspace.from_config(auth = auth)\n",
"```\n",
"For more details, see aka.ms/aml-notebook-auth"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"ws = Workspace.from_config()\n",
"dstor = ws.get_default_datastore()\n",
"\n",
"# Choose a name for the run history container in the workspace.\n",
"experiment_name = \"retrain-noaaweather\"\n",
"experiment = Experiment(ws, experiment_name)\n",
"\n",
"output = {}\n",
"output[\"Subscription ID\"] = ws.subscription_id\n",
"output[\"Workspace\"] = ws.name\n",
"output[\"Resource Group\"] = ws.resource_group\n",
"output[\"Location\"] = ws.location\n",
"output[\"Run History Name\"] = experiment_name\n",
"pd.set_option(\"display.max_colwidth\", None)\n",
"outputDf = pd.DataFrame(data=output, index=[\"\"])\n",
"outputDf.T"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Compute \n",
"\n",
"#### Create or Attach existing AmlCompute\n",
"\n",
"You will need to create a compute target for your AutoML run. In this tutorial, you create AmlCompute as your training compute resource.\n",
"\n",
"> Note that if you have an AzureML Data Scientist role, you will not have permission to create compute resources. Talk to your workspace or IT admin to create the compute targets described in this section, if they do not already exist.\n",
"\n",
"#### Creation of AmlCompute takes approximately 5 minutes. \n",
"If the AmlCompute with that name is already in your workspace this code will skip the creation process.\n",
"As with other Azure services, there are limits on certain resources (e.g. AmlCompute) associated with the Azure Machine Learning service. Please read [this article](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-manage-quotas) on the default limits and how to request more quota."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core.compute import ComputeTarget, AmlCompute\n",
"from azureml.core.compute_target import ComputeTargetException\n",
"\n",
"# Choose a name for your CPU cluster\n",
"amlcompute_cluster_name = \"cont-cluster\"\n",
"\n",
"# Verify that cluster does not exist already\n",
"try:\n",
" compute_target = ComputeTarget(workspace=ws, name=amlcompute_cluster_name)\n",
" print(\"Found existing cluster, use it.\")\n",
"except ComputeTargetException:\n",
" compute_config = AmlCompute.provisioning_configuration(\n",
" vm_size=\"STANDARD_DS12_V2\", max_nodes=4\n",
" )\n",
" compute_target = ComputeTarget.create(ws, amlcompute_cluster_name, compute_config)\n",
"compute_target.wait_for_completion(show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Run Configuration"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core.runconfig import CondaDependencies, RunConfiguration\n",
"\n",
"# create a new RunConfig object\n",
"conda_run_config = RunConfiguration(framework=\"python\")\n",
"\n",
"# Set compute target to AmlCompute\n",
"conda_run_config.target = compute_target\n",
"\n",
"conda_run_config.environment.docker.enabled = True\n",
"\n",
"cd = CondaDependencies.create(\n",
" pip_packages=[\n",
" \"azureml-sdk[automl]\",\n",
" \"applicationinsights\",\n",
" \"azureml-opendatasets\",\n",
" \"azureml-defaults\",\n",
" ],\n",
" conda_packages=[\"numpy==1.16.2\"],\n",
" pin_sdk_version=False,\n",
")\n",
"conda_run_config.environment.python.conda_dependencies = cd\n",
"\n",
"print(\"run config is ready\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Data Ingestion Pipeline \n",
"For this demo, we will use NOAA weather data from [Azure Open Datasets](https://azure.microsoft.com/services/open-datasets/). You can replace this with your own dataset, or you can skip this pipeline if you already have a time-series based `TabularDataset`.\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# The name and target column of the Dataset to create\n",
"dataset = \"NOAA-Weather-DS4\"\n",
"target_column_name = \"temperature\""
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"\n",
"### Upload Data Step\n",
"The data ingestion pipeline has a single step with a script to query the latest weather data and upload it to the blob store. During the first run, the script will create and register a time-series based `TabularDataset` with the past one week of weather data. For each subsequent run, the script will create a partition in the blob store by querying NOAA for new weather data since the last modified time of the dataset (`dataset.data_changed_time`) and creating a data.csv file."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.pipeline.core import Pipeline, PipelineParameter\n",
"from azureml.pipeline.steps import PythonScriptStep\n",
"\n",
"ds_name = PipelineParameter(name=\"ds_name\", default_value=dataset)\n",
"upload_data_step = PythonScriptStep(\n",
" script_name=\"upload_weather_data.py\",\n",
" allow_reuse=False,\n",
" name=\"upload_weather_data\",\n",
" arguments=[\"--ds_name\", ds_name],\n",
" compute_target=compute_target,\n",
" runconfig=conda_run_config,\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Submit Pipeline Run"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"data_pipeline = Pipeline(\n",
" description=\"pipeline_with_uploaddata\", workspace=ws, steps=[upload_data_step]\n",
")\n",
"data_pipeline_run = experiment.submit(\n",
" data_pipeline, pipeline_parameters={\"ds_name\": dataset}\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"data_pipeline_run.wait_for_completion(show_output=False)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Training Pipeline\n",
"### Prepare Training Data Step\n",
"\n",
"Script to check if new data is available since the model was last trained. If no new data is available, we cancel the remaining pipeline steps. We need to set allow_reuse flag to False to allow the pipeline to run even when inputs don't change. We also need the name of the model to check the time the model was last trained."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.pipeline.core import PipelineData\n",
"\n",
"# The model name with which to register the trained model in the workspace.\n",
"model_name = PipelineParameter(\"model_name\", default_value=\"noaaweatherds\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"data_prep_step = PythonScriptStep(\n",
" script_name=\"check_data.py\",\n",
" allow_reuse=False,\n",
" name=\"check_data\",\n",
" arguments=[\"--ds_name\", ds_name, \"--model_name\", model_name],\n",
" compute_target=compute_target,\n",
" runconfig=conda_run_config,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core import Dataset\n",
"\n",
"train_ds = Dataset.get_by_name(ws, dataset)\n",
"train_ds = train_ds.drop_columns([\"partition_date\"])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### AutoMLStep\n",
"Create an AutoMLConfig and a training step."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.train.automl import AutoMLConfig\n",
"from azureml.pipeline.steps import AutoMLStep\n",
"\n",
"automl_settings = {\n",
" \"iteration_timeout_minutes\": 10,\n",
" \"experiment_timeout_hours\": 0.25,\n",
" \"n_cross_validations\": 3,\n",
" \"primary_metric\": \"r2_score\",\n",
" \"max_concurrent_iterations\": 3,\n",
" \"max_cores_per_iteration\": -1,\n",
" \"verbosity\": logging.INFO,\n",
" \"enable_early_stopping\": True,\n",
"}\n",
"\n",
"automl_config = AutoMLConfig(\n",
" task=\"regression\",\n",
" debug_log=\"automl_errors.log\",\n",
" path=\".\",\n",
" compute_target=compute_target,\n",
" training_data=train_ds,\n",
" label_column_name=target_column_name,\n",
" **automl_settings,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.pipeline.core import PipelineData, TrainingOutput\n",
"\n",
"metrics_output_name = \"metrics_output\"\n",
"best_model_output_name = \"best_model_output\"\n",
"\n",
"metrics_data = PipelineData(\n",
" name=\"metrics_data\",\n",
" datastore=dstor,\n",
" pipeline_output_name=metrics_output_name,\n",
" training_output=TrainingOutput(type=\"Metrics\"),\n",
")\n",
"model_data = PipelineData(\n",
" name=\"model_data\",\n",
" datastore=dstor,\n",
" pipeline_output_name=best_model_output_name,\n",
" training_output=TrainingOutput(type=\"Model\"),\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"automl_step = AutoMLStep(\n",
" name=\"automl_module\",\n",
" automl_config=automl_config,\n",
" outputs=[metrics_data, model_data],\n",
" allow_reuse=False,\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Register Model Step\n",
"Script to register the model to the workspace. "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"register_model_step = PythonScriptStep(\n",
" script_name=\"register_model.py\",\n",
" name=\"register_model\",\n",
" allow_reuse=False,\n",
" arguments=[\n",
" \"--model_name\",\n",
" model_name,\n",
" \"--model_path\",\n",
" model_data,\n",
" \"--ds_name\",\n",
" ds_name,\n",
" ],\n",
" inputs=[model_data],\n",
" compute_target=compute_target,\n",
" runconfig=conda_run_config,\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Submit Pipeline Run"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"training_pipeline = Pipeline(\n",
" description=\"training_pipeline\",\n",
" workspace=ws,\n",
" steps=[data_prep_step, automl_step, register_model_step],\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"training_pipeline_run = experiment.submit(\n",
" training_pipeline,\n",
" pipeline_parameters={\"ds_name\": dataset, \"model_name\": \"noaaweatherds\"},\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"training_pipeline_run.wait_for_completion(show_output=False)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Publish Retraining Pipeline and Schedule\n",
"Once we are happy with the pipeline, we can publish the training pipeline to the workspace and create a schedule to trigger on blob change. The schedule polls the blob store where the data is being uploaded and runs the retraining pipeline if there is a data change. A new version of the model will be registered to the workspace once the run is complete."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"pipeline_name = \"Retraining-Pipeline-NOAAWeather\"\n",
"\n",
"published_pipeline = training_pipeline.publish(\n",
" name=pipeline_name, description=\"Pipeline that retrains AutoML model\"\n",
")\n",
"\n",
"published_pipeline"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.pipeline.core import Schedule\n",
"\n",
"schedule = Schedule.create(\n",
" workspace=ws,\n",
" name=\"RetrainingSchedule\",\n",
" pipeline_parameters={\"ds_name\": dataset, \"model_name\": \"noaaweatherds\"},\n",
" pipeline_id=published_pipeline.id,\n",
" experiment_name=experiment_name,\n",
" datastore=dstor,\n",
" wait_for_provisioning=True,\n",
" polling_interval=1440,\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Test Retraining\n",
"Here we setup the data ingestion pipeline to run on a schedule, to verify that the retraining pipeline runs as expected. \n",
"\n",
"Note: \n",
"* Azure NOAA Weather data is updated daily and retraining will not trigger if there is no new data available. \n",
"* Depending on the polling interval set in the schedule, the retraining may take some time trigger after data ingestion pipeline completes."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"pipeline_name = \"DataIngestion-Pipeline-NOAAWeather\"\n",
"\n",
"published_pipeline = training_pipeline.publish(\n",
" name=pipeline_name, description=\"Pipeline that updates NOAAWeather Dataset\"\n",
")\n",
"\n",
"published_pipeline"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.pipeline.core import Schedule\n",
"\n",
"schedule = Schedule.create(\n",
" workspace=ws,\n",
" name=\"RetrainingSchedule-DataIngestion\",\n",
" pipeline_parameters={\"ds_name\": dataset},\n",
" pipeline_id=published_pipeline.id,\n",
" experiment_name=experiment_name,\n",
" datastore=dstor,\n",
" wait_for_provisioning=True,\n",
" polling_interval=1440,\n",
")"
]
}
],
"metadata": {
"authors": [
{
"name": "vivijay"
}
],
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.6"
}
},
"nbformat": 4,
"nbformat_minor": 2
}

4
how-to-use-azureml/automated-machine-learning/continuous-retraining/auto-ml-continuous-retraining.yml Normal file
View File

@@ -0,0 +1,4 @@
name: auto-ml-continuous-retraining
dependencies:
- pip:
- azureml-sdk

46
how-to-use-azureml/automated-machine-learning/continuous-retraining/check_data.py Normal file
View File

@@ -0,0 +1,46 @@
import argparse
import os
import azureml.core
from datetime import datetime
import pandas as pd
import pytz
from azureml.core import Dataset, Model
from azureml.core.run import Run, _OfflineRun
from azureml.core import Workspace
run = Run.get_context()
ws = None
if type(run) == _OfflineRun:
ws = Workspace.from_config()
else:
ws = run.experiment.workspace
print("Check for new data.")
parser = argparse.ArgumentParser("split")
parser.add_argument("--ds_name", help="input dataset name")
parser.add_argument("--model_name", help="name of the deployed model")
args = parser.parse_args()
print("Argument 1(ds_name): %s" % args.ds_name)
print("Argument 2(model_name): %s" % args.model_name)
# Get the latest registered model
try:
model = Model(ws, args.model_name)
last_train_time = model.created_time
print("Model was last trained on {0}.".format(last_train_time))
except Exception as e:
print("Could not get last model train time.")
last_train_time = datetime.min.replace(tzinfo=pytz.UTC)
train_ds = Dataset.get_by_name(ws, args.ds_name)
dataset_changed_time = train_ds.data_changed_time
if not dataset_changed_time > last_train_time:
print("Cancelling run since there is no new data.")
run.parent.cancel()
else:
# New data is available since the model was last trained
print("Dataset was last updated on {0}. Retraining...".format(dataset_changed_time))

35
how-to-use-azureml/automated-machine-learning/continuous-retraining/register_model.py Normal file
View File

@@ -0,0 +1,35 @@
from azureml.core.model import Model, Dataset
from azureml.core.run import Run, _OfflineRun
from azureml.core import Workspace
import argparse
parser = argparse.ArgumentParser()
parser.add_argument("--model_name")
parser.add_argument("--model_path")
parser.add_argument("--ds_name")
args = parser.parse_args()
print("Argument 1(model_name): %s" % args.model_name)
print("Argument 2(model_path): %s" % args.model_path)
print("Argument 3(ds_name): %s" % args.ds_name)
run = Run.get_context()
ws = None
if type(run) == _OfflineRun:
ws = Workspace.from_config()
else:
ws = run.experiment.workspace
train_ds = Dataset.get_by_name(ws, args.ds_name)
datasets = [(Dataset.Scenario.TRAINING, train_ds)]
# Register model with training dataset
model = Model.register(
workspace=ws,
model_path=args.model_path,
model_name=args.model_name,
datasets=datasets,
)
print("Registered version {0} of model {1}".format(model.version, model.name))

157
how-to-use-azureml/automated-machine-learning/continuous-retraining/upload_weather_data.py Normal file
View File

@@ -0,0 +1,157 @@
import argparse
import os
from datetime import datetime
from dateutil.relativedelta import relativedelta
import pandas as pd
import traceback
from azureml.core import Dataset
from azureml.core.run import Run, _OfflineRun
from azureml.core import Workspace
from azureml.opendatasets import NoaaIsdWeather
run = Run.get_context()
ws = None
if type(run) == _OfflineRun:
ws = Workspace.from_config()
else:
ws = run.experiment.workspace
usaf_list = [
"725724",
"722149",
"723090",
"722159",
"723910",
"720279",
"725513",
"725254",
"726430",
"720381",
"723074",
"726682",
"725486",
"727883",
"723177",
"722075",
"723086",
"724053",
"725070",
"722073",
"726060",
"725224",
"725260",
"724520",
"720305",
"724020",
"726510",
"725126",
"722523",
"703333",
"722249",
"722728",
"725483",
"722972",
"724975",
"742079",
"727468",
"722193",
"725624",
"722030",
"726380",
"720309",
"722071",
"720326",
"725415",
"724504",
"725665",
"725424",
"725066",
]
def get_noaa_data(start_time, end_time):
columns = [
"usaf",
"wban",
"datetime",
"latitude",
"longitude",
"elevation",
"windAngle",
"windSpeed",
"temperature",
"stationName",
"p_k",
]
isd = NoaaIsdWeather(start_time, end_time, cols=columns)
noaa_df = isd.to_pandas_dataframe()
df_filtered = noaa_df[noaa_df["usaf"].isin(usaf_list)]
df_filtered.reset_index(drop=True)
print(
"Received {0} rows of training data between {1} and {2}".format(
df_filtered.shape[0], start_time, end_time
)
)
return df_filtered
print("Check for new data and prepare the data")
parser = argparse.ArgumentParser("split")
parser.add_argument("--ds_name", help="name of the Dataset to update")
args = parser.parse_args()
print("Argument 1(ds_name): %s" % args.ds_name)
dstor = ws.get_default_datastore()
register_dataset = False
end_time = datetime.utcnow()
try:
ds = Dataset.get_by_name(ws, args.ds_name)
end_time_last_slice = ds.data_changed_time.replace(tzinfo=None)
print("Dataset {0} last updated on {1}".format(args.ds_name, end_time_last_slice))
except Exception:
print(traceback.format_exc())
print(
"Dataset with name {0} not found, registering new dataset.".format(args.ds_name)
)
register_dataset = True
end_time = datetime(2021, 5, 1, 0, 0)
end_time_last_slice = end_time - relativedelta(weeks=2)
train_df = get_noaa_data(end_time_last_slice, end_time)
if train_df.size > 0:
print(
"Received {0} rows of new data after {1}.".format(
train_df.shape[0], end_time_last_slice
)
)
folder_name = "{}/{:04d}/{:02d}/{:02d}/{:02d}/{:02d}/{:02d}".format(
args.ds_name,
end_time.year,
end_time.month,
end_time.day,
end_time.hour,
end_time.minute,
end_time.second,
)
file_path = "{0}/data.csv".format(folder_name)
# Add a new partition to the registered dataset
os.makedirs(folder_name, exist_ok=True)
train_df.to_csv(file_path, index=False)
dstor.upload_files(
files=[file_path], target_path=folder_name, overwrite=True, show_progress=True
)
else:
print("No new data since {0}.".format(end_time_last_slice))
if register_dataset:
ds = Dataset.Tabular.from_delimited_files(
dstor.path("{}/**/*.csv".format(args.ds_name)),
partition_format="/{partition_date:yyyy/MM/dd/HH/mm/ss}/data.csv",
)
ds.register(ws, name=args.ds_name)

92
how-to-use-azureml/automated-machine-learning/experimental/README.md Normal file
View File

@@ -0,0 +1,92 @@
# Experimental Notebooks for Automated ML
Notebooks listed in this folder are leveraging experimental features. Namespaces or function signitures may change in future SDK releases. The notebooks published here will reflect the latest supported APIs. All of these notebooks can run on a client-only installation of the Automated ML SDK.
The client only installation doesn't contain any of the machine learning libraries, such as scikit-learn, xgboost, or tensorflow, making it much faster to install and is less likely to conflict with any packages in an existing environment. However, since the ML libraries are not available locally, models cannot be downloaded and loaded directly in the client. To replace the functionality of having models locally, these notebooks also demonstrate the ModelProxy feature which will allow you to submit a predict/forecast to the training environment.
<a name="localconda"></a>
## Setup using a Local Conda environment
To run these notebook on your own notebook server, use these installation instructions.
The instructions below will install everything you need and then start a Jupyter notebook.
If you would like to use a lighter-weight version of the client that does not install all of the machine learning libraries locally, you can leverage the [experimental notebooks.](experimental/README.md)
### 1. Install mini-conda from [here](https://conda.io/miniconda.html), choose 64-bit Python 3.7 or higher.
- **Note**: if you already have conda installed, you can keep using it but it should be version 4.4.10 or later (as shown by: conda -V). If you have a previous version installed, you can update it using the command: conda update conda.
There's no need to install mini-conda specifically.
### 2. Downloading the sample notebooks
- Download the sample notebooks from [GitHub](https://github.com/Azure/MachineLearningNotebooks) as zip and extract the contents to a local directory. The automated ML sample notebooks are in the "automated-machine-learning" folder.
### 3. Setup a new conda environment
The **automl_setup_thin_client** script creates a new conda environment, installs the necessary packages, configures the widget and starts a jupyter notebook. It takes the conda environment name as an optional parameter. The default conda environment name is azure_automl_experimental. The exact command depends on the operating system. See the specific sections below for Windows, Mac and Linux. It can take about 10 minutes to execute.
Packages installed by the **automl_setup** script:
<ul><li>python</li><li>nb_conda</li><li>matplotlib</li><li>numpy</li><li>cython</li><li>urllib3</li><li>pandas</li><li>azureml-sdk</li><li>azureml-widgets</li><li>pandas-ml</li></ul>
For more details refer to the [automl_env_thin_client.yml](./automl_env_thin_client.yml)
## Windows
Start an **Anaconda Prompt** window, cd to the **how-to-use-azureml/automated-machine-learning/experimental** folder where the sample notebooks were extracted and then run:
```
automl_setup_thin_client
```
## Mac
Install "Command line developer tools" if it is not already installed (you can use the command: `xcode-select --install`).
Start a Terminal windows, cd to the **how-to-use-azureml/automated-machine-learning/experimental** folder where the sample notebooks were extracted and then run:
```
bash automl_setup_thin_client_mac.sh
```
## Linux
cd to the **how-to-use-azureml/automated-machine-learning/experimental** folder where the sample notebooks were extracted and then run:
```
bash automl_setup_thin_client_linux.sh
```
### 4. Running configuration.ipynb
- Before running any samples you next need to run the configuration notebook. Click on [configuration](../../configuration.ipynb) notebook
- Execute the cells in the notebook to Register Machine Learning Services Resource Provider and create a workspace. (*instructions in notebook*)
### 5. Running Samples
- Please make sure you use the Python [conda env:azure_automl_experimental] kernel when trying the sample Notebooks.
- Follow the instructions in the individual notebooks to explore various features in automated ML.
### 6. Starting jupyter notebook manually
To start your Jupyter notebook manually, use:
```
conda activate azure_automl
jupyter notebook
```
or on Mac or Linux:
```
source activate azure_automl
jupyter notebook
```
<a name="samples"></a>
# Automated ML SDK Sample Notebooks
- [auto-ml-regression-model-proxy.ipynb](regression-model-proxy/auto-ml-regression-model-proxy.ipynb)
- Dataset: Hardware Performance Dataset
- Simple example of using automated ML for regression
- Uses azure compute for training
- Uses ModelProxy for submitting prediction to training environment on azure compute
<a name="documentation"></a>
See [Configure automated machine learning experiments](https://docs.microsoft.com/azure/machine-learning/service/how-to-configure-auto-train) to learn how more about the the settings and features available for automated machine learning experiments.
<a name="pythoncommand"></a>
# Running using python command
Jupyter notebook provides a File / Download as / Python (.py) option for saving the notebook as a Python file.
You can then run this file using the python command.
However, on Windows the file needs to be modified before it can be run.
The following condition must be added to the main code in the file:
if __name__ == "__main__":
The main code of the file must be indented so that it is under this condition.

63
how-to-use-azureml/automated-machine-learning/experimental/automl_setup_thin_client.cmd Normal file
View File

@@ -0,0 +1,63 @@
@echo off
set conda_env_name=%1
set automl_env_file=%2
set options=%3
set PIP_NO_WARN_SCRIPT_LOCATION=0
IF "%conda_env_name%"=="" SET conda_env_name="azure_automl_experimental"
IF "%automl_env_file%"=="" SET automl_env_file="automl_thin_client_env.yml"
IF NOT EXIST %automl_env_file% GOTO YmlMissing
IF "%CONDA_EXE%"=="" GOTO CondaMissing
call conda activate %conda_env_name% 2>nul:
if not errorlevel 1 (
echo Upgrading existing conda environment %conda_env_name%
call pip uninstall azureml-train-automl -y -q
call conda env update --name %conda_env_name% --file %automl_env_file%
if errorlevel 1 goto ErrorExit
) else (
call conda env create -f %automl_env_file% -n %conda_env_name%
)
call conda activate %conda_env_name% 2>nul:
if errorlevel 1 goto ErrorExit
call python -m ipykernel install --user --name %conda_env_name% --display-name "Python (%conda_env_name%)"
REM azureml.widgets is now installed as part of the pip install under the conda env.
REM Removing the old user install so that the notebooks will use the latest widget.
call jupyter nbextension uninstall --user --py azureml.widgets
echo.
echo.
echo ***************************************
echo * AutoML setup completed successfully *
echo ***************************************
IF NOT "%options%"=="nolaunch" (
echo.
echo Starting jupyter notebook - please run the configuration notebook
echo.
jupyter notebook --log-level=50 --notebook-dir='..\..'
)
goto End
:CondaMissing
echo Please run this script from an Anaconda Prompt window.
echo You can start an Anaconda Prompt window by
echo typing Anaconda Prompt on the Start menu.
echo If you don't see the Anaconda Prompt app, install Miniconda.
echo If you are running an older version of Miniconda or Anaconda,
echo you can upgrade using the command: conda update conda
goto End
:YmlMissing
echo File %automl_env_file% not found.
:ErrorExit
echo Install failed
:End

53
how-to-use-azureml/automated-machine-learning/experimental/automl_setup_thin_client_linux.sh Normal file
View File

@@ -0,0 +1,53 @@
#!/bin/bash
CONDA_ENV_NAME=$1
AUTOML_ENV_FILE=$2
OPTIONS=$3
PIP_NO_WARN_SCRIPT_LOCATION=0
if [ "$CONDA_ENV_NAME" == "" ]
then
CONDA_ENV_NAME="azure_automl_experimental"
fi
if [ "$AUTOML_ENV_FILE" == "" ]
then
AUTOML_ENV_FILE="automl_thin_client_env.yml"
fi
if [ ! -f $AUTOML_ENV_FILE ]; then
echo "File $AUTOML_ENV_FILE not found"
exit 1
fi
if source activate $CONDA_ENV_NAME 2> /dev/null
then
echo "Upgrading existing conda environment" $CONDA_ENV_NAME
pip uninstall azureml-train-automl -y -q
conda env update --name $CONDA_ENV_NAME --file $AUTOML_ENV_FILE &&
jupyter nbextension uninstall --user --py azureml.widgets
else
conda env create -f $AUTOML_ENV_FILE -n $CONDA_ENV_NAME &&
source activate $CONDA_ENV_NAME &&
python -m ipykernel install --user --name $CONDA_ENV_NAME --display-name "Python ($CONDA_ENV_NAME)" &&
jupyter nbextension uninstall --user --py azureml.widgets &&
echo "" &&
echo "" &&
echo "***************************************" &&
echo "* AutoML setup completed successfully *" &&
echo "***************************************" &&
if [ "$OPTIONS" != "nolaunch" ]
then
echo "" &&
echo "Starting jupyter notebook - please run the configuration notebook" &&
echo "" &&
jupyter notebook --log-level=50 --notebook-dir '../..'
fi
fi
if [ $? -gt 0 ]
then
echo "Installation failed"
fi

55
how-to-use-azureml/automated-machine-learning/experimental/automl_setup_thin_client_mac.sh Normal file
View File

@@ -0,0 +1,55 @@
#!/bin/bash
CONDA_ENV_NAME=$1
AUTOML_ENV_FILE=$2
OPTIONS=$3
PIP_NO_WARN_SCRIPT_LOCATION=0
if [ "$CONDA_ENV_NAME" == "" ]
then
CONDA_ENV_NAME="azure_automl_experimental"
fi
if [ "$AUTOML_ENV_FILE" == "" ]
then
AUTOML_ENV_FILE="automl_thin_client_env_mac.yml"
fi
if [ ! -f $AUTOML_ENV_FILE ]; then
echo "File $AUTOML_ENV_FILE not found"
exit 1
fi
if source activate $CONDA_ENV_NAME 2> /dev/null
then
echo "Upgrading existing conda environment" $CONDA_ENV_NAME
pip uninstall azureml-train-automl -y -q
conda env update --name $CONDA_ENV_NAME --file $AUTOML_ENV_FILE &&
jupyter nbextension uninstall --user --py azureml.widgets
else
conda env create -f $AUTOML_ENV_FILE -n $CONDA_ENV_NAME &&
source activate $CONDA_ENV_NAME &&
conda install lightgbm -c conda-forge -y &&
python -m ipykernel install --user --name $CONDA_ENV_NAME --display-name "Python ($CONDA_ENV_NAME)" &&
jupyter nbextension uninstall --user --py azureml.widgets &&
echo "" &&
echo "" &&
echo "***************************************" &&
echo "* AutoML setup completed successfully *" &&
echo "***************************************" &&
if [ "$OPTIONS" != "nolaunch" ]
then
echo "" &&
echo "Starting jupyter notebook - please run the configuration notebook" &&
echo "" &&
jupyter notebook --log-level=50 --notebook-dir '../..'
fi
fi
if [ $? -gt 0 ]
then
echo "Installation failed"
fi

17
how-to-use-azureml/automated-machine-learning/experimental/automl_thin_client_env.yml Normal file
View File

@@ -0,0 +1,17 @@
name: azure_automl_experimental
dependencies:
# The python interpreter version.
# Currently Azure ML only supports 3.6.0 and later.
- pip<=20.2.4
- python>=3.6.0,<3.9
- cython==0.29.14
- urllib3==1.26.7
- PyJWT < 2.0.0
- numpy==1.18.5
- pip:
# Required packages for AzureML execution, history, and data preparation.
- azureml-defaults
- azureml-sdk
- azureml-widgets
- pandas

20
how-to-use-azureml/automated-machine-learning/experimental/automl_thin_client_env_mac.yml Normal file
View File

@@ -0,0 +1,20 @@
name: azure_automl_experimental
channels:
- conda-forge
- main
dependencies:
# The python interpreter version.
# Currently Azure ML only supports 3.6.0 and later.
- pip<=20.2.4
- nomkl
- python>=3.6.0,<3.9
- urllib3==1.26.7
- PyJWT < 2.0.0
- numpy==1.19.5
- pip:
# Required packages for AzureML execution, history, and data preparation.
- azureml-defaults
- azureml-sdk
- azureml-widgets
- pandas

420
how-to-use-azureml/automated-machine-learning/experimental/classification-credit-card-fraud-local-managed/auto-ml-classification-credit-card-fraud-local-managed.ipynb Normal file
View File

@@ -0,0 +1,420 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copyright (c) Microsoft Corporation. All rights reserved.\n",
"\n",
"Licensed under the MIT License."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"![Impressions](https://PixelServer20190423114238.azurewebsites.net/api/impressions/MachineLearningNotebooks/how-to-use-azureml/automated-machine-learning/experimental/classification-credit-card-fraud/auto-ml-classification-credit-card-fraud.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Automated Machine Learning\n",
"_**Classification of credit card fraudulent transactions on local managed compute **_\n",
"\n",
"## Contents\n",
"1. [Introduction](#Introduction)\n",
"1. [Setup](#Setup)\n",
"1. [Train](#Train)\n",
"1. [Results](#Results)\n",
"1. [Test](#Test)\n",
"1. [Acknowledgements](#Acknowledgements)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Introduction\n",
"\n",
"In this example we use the associated credit card dataset to showcase how you can use AutoML for a simple classification problem. The goal is to predict if a credit card transaction is considered a fraudulent charge.\n",
"\n",
"This notebook is using local managed compute to train the model.\n",
"\n",
"If you are using an Azure Machine Learning Compute Instance, you are all set. Otherwise, go through the [configuration](../../../configuration.ipynb) notebook first if you haven't already to establish your connection to the AzureML Workspace. \n",
"\n",
"In this notebook you will learn how to:\n",
"1. Create an experiment using an existing workspace.\n",
"2. Configure AutoML using `AutoMLConfig`.\n",
"3. Train the model using local managed compute.\n",
"4. Explore the results.\n",
"5. Test the fitted model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Setup\n",
"\n",
"As part of the setup you have already created an Azure ML `Workspace` object. For Automated ML you will need to create an `Experiment` object, which is a named object in a `Workspace` used to run experiments."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import logging\n",
"\n",
"import pandas as pd\n",
"\n",
"import azureml.core\n",
"from azureml.core.compute_target import LocalTarget\n",
"from azureml.core.experiment import Experiment\n",
"from azureml.core.workspace import Workspace\n",
"from azureml.core.dataset import Dataset\n",
"from azureml.train.automl import AutoMLConfig"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This sample notebook may use features that are not available in previous versions of the Azure ML SDK."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"print(\"This notebook was created using version 1.40.0 of the Azure ML SDK\")\n",
"print(\"You are currently using version\", azureml.core.VERSION, \"of the Azure ML SDK\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"ws = Workspace.from_config()\n",
"\n",
"# choose a name for experiment\n",
"experiment_name = 'automl-local-managed'\n",
"\n",
"experiment=Experiment(ws, experiment_name)\n",
"\n",
"output = {}\n",
"output['Subscription ID'] = ws.subscription_id\n",
"output['Workspace'] = ws.name\n",
"output['Resource Group'] = ws.resource_group\n",
"output['Location'] = ws.location\n",
"output['Experiment Name'] = experiment.name\n",
"pd.set_option('display.max_colwidth', None)\n",
"outputDf = pd.DataFrame(data = output, index = [''])\n",
"outputDf.T"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Determine if local docker is configured for Linux images\n",
"\n",
"Local managed runs will leverage a Linux docker container to submit the run to. Due to this, the docker needs to be configured to use Linux containers."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# Check if Docker is installed and Linux containers are enabled\n",
"import subprocess\n",
"from subprocess import CalledProcessError\n",
"try:\n",
" assert subprocess.run(\"docker -v\", shell=True).returncode == 0, 'Local Managed runs require docker to be installed.'\n",
" out = subprocess.check_output(\"docker system info\", shell=True).decode('ascii')\n",
" assert \"OSType: linux\" in out, 'Docker engine needs to be configured to use Linux containers.' \\\n",
" 'https://docs.docker.com/docker-for-windows/#switch-between-windows-and-linux-containers'\n",
"except CalledProcessError as ex:\n",
" raise Exception('Local Managed runs require docker to be installed.') from ex"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Data"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Load Data\n",
"\n",
"Load the credit card dataset from a csv file containing both training features and labels. The features are inputs to the model, while the training labels represent the expected output of the model. Next, we'll split the data using random_split and extract the training data for the model."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"data = \"https://automlsamplenotebookdata.blob.core.windows.net/automl-sample-notebook-data/creditcard.csv\"\n",
"dataset = Dataset.Tabular.from_delimited_files(data)\n",
"training_data, validation_data = dataset.random_split(percentage=0.8, seed=223)\n",
"label_column_name = 'Class'"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Train\n",
"\n",
"Instantiate a AutoMLConfig object. This defines the settings and data used to run the experiment.\n",
"\n",
"|Property|Description|\n",
"|-|-|\n",
"|**task**|classification or regression|\n",
"|**primary_metric**|This is the metric that you want to optimize. Classification supports the following primary metrics: <br><i>accuracy</i><br><i>AUC_weighted</i><br><i>average_precision_score_weighted</i><br><i>norm_macro_recall</i><br><i>precision_score_weighted</i>|\n",
"|**enable_early_stopping**|Stop the run if the metric score is not showing improvement.|\n",
"|**n_cross_validations**|Number of cross validation splits.|\n",
"|**training_data**|Input dataset, containing both features and label column.|\n",
"|**label_column_name**|The name of the label column.|\n",
"|**enable_local_managed**|Enable the experimental local-managed scenario.|\n",
"\n",
"**_You can find more information about primary metrics_** [here](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-configure-auto-train#primary-metric)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"automl_settings = {\n",
" \"n_cross_validations\": 3,\n",
" \"primary_metric\": 'average_precision_score_weighted',\n",
" \"enable_early_stopping\": True,\n",
" \"experiment_timeout_hours\": 0.3, #for real scenarios we recommend a timeout of at least one hour \n",
" \"verbosity\": logging.INFO,\n",
"}\n",
"\n",
"automl_config = AutoMLConfig(task = 'classification',\n",
" debug_log = 'automl_errors.log',\n",
" compute_target = LocalTarget(),\n",
" enable_local_managed = True,\n",
" training_data = training_data,\n",
" label_column_name = label_column_name,\n",
" **automl_settings\n",
" )"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Call the `submit` method on the experiment object and pass the run configuration. Depending on the data and the number of iterations this can run for a while. Validation errors and current status will be shown when setting `show_output=True` and the execution will be synchronous."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"parent_run = experiment.submit(automl_config, show_output = True)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# If you need to retrieve a run that already started, use the following code\n",
"#from azureml.train.automl.run import AutoMLRun\n",
"#parent_run = AutoMLRun(experiment = experiment, run_id = '<replace with your run id>')"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"parent_run"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Results"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Explain model\n",
"\n",
"Automated ML models can be explained and visualized using the SDK Explainability library. "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Analyze results\n",
"\n",
"### Retrieve the Best Child Run\n",
"\n",
"Below we select the best pipeline from our iterations. The `get_best_child` method returns the best run. Overloads on `get_best_child` allow you to retrieve the best run for *any* logged metric."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"best_run = parent_run.get_best_child()\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Test the fitted model\n",
"\n",
"Now that the model is trained, split the data in the same way the data was split for training (The difference here is the data is being split locally) and then run the test data through the trained model to get the predicted values."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"X_test_df = validation_data.drop_columns(columns=[label_column_name])\n",
"y_test_df = validation_data.keep_columns(columns=[label_column_name], validate=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Creating ModelProxy for submitting prediction runs to the training environment.\n",
"We will create a ModelProxy for the best child run, which will allow us to submit a run that does the prediction in the training environment. Unlike the local client, which can have different versions of some libraries, the training environment will have all the compatible libraries for the model already."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.train.automl.model_proxy import ModelProxy\n",
"best_model_proxy = ModelProxy(best_run)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# call the predict functions on the model proxy\n",
"y_pred = best_model_proxy.predict(X_test_df).to_pandas_dataframe()\n",
"y_pred"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Acknowledgements"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This Credit Card fraud Detection dataset is made available under the Open Database License: http://opendatacommons.org/licenses/odbl/1.0/. Any rights in individual contents of the database are licensed under the Database Contents License: http://opendatacommons.org/licenses/dbcl/1.0/ and is available at: https://www.kaggle.com/mlg-ulb/creditcardfraud\n",
"\n",
"\n",
"The dataset has been collected and analysed during a research collaboration of Worldline and the Machine Learning Group (http://mlg.ulb.ac.be) of ULB (Universit\u00c3\u0192\u00c2\u00a9 Libre de Bruxelles) on big data mining and fraud detection. More details on current and past projects on related topics are available on https://www.researchgate.net/project/Fraud-detection-5 and the page of the DefeatFraud project\n",
"Please cite the following works: \n",
"\u00c3\u00a2\u00e2\u201a\u00ac\u00c2\u00a2\tAndrea Dal Pozzolo, Olivier Caelen, Reid A. Johnson and Gianluca Bontempi. Calibrating Probability with Undersampling for Unbalanced Classification. In Symposium on Computational Intelligence and Data Mining (CIDM), IEEE, 2015\n",
"\u00c3\u00a2\u00e2\u201a\u00ac\u00c2\u00a2\tDal Pozzolo, Andrea; Caelen, Olivier; Le Borgne, Yann-Ael; Waterschoot, Serge; Bontempi, Gianluca. Learned lessons in credit card fraud detection from a practitioner perspective, Expert systems with applications,41,10,4915-4928,2014, Pergamon\n",
"\u00c3\u00a2\u00e2\u201a\u00ac\u00c2\u00a2\tDal Pozzolo, Andrea; Boracchi, Giacomo; Caelen, Olivier; Alippi, Cesare; Bontempi, Gianluca. Credit card fraud detection: a realistic modeling and a novel learning strategy, IEEE transactions on neural networks and learning systems,29,8,3784-3797,2018,IEEE\n",
"o\tDal Pozzolo, Andrea Adaptive Machine learning for credit card fraud detection ULB MLG PhD thesis (supervised by G. Bontempi)\n",
"\u00c3\u00a2\u00e2\u201a\u00ac\u00c2\u00a2\tCarcillo, Fabrizio; Dal Pozzolo, Andrea; Le Borgne, Yann-A\u00c3\u0192\u00c2\u00abl; Caelen, Olivier; Mazzer, Yannis; Bontempi, Gianluca. Scarff: a scalable framework for streaming credit card fraud detection with Spark, Information fusion,41, 182-194,2018,Elsevier\n",
"\u00c3\u00a2\u00e2\u201a\u00ac\u00c2\u00a2\tCarcillo, Fabrizio; Le Borgne, Yann-A\u00c3\u0192\u00c2\u00abl; Caelen, Olivier; Bontempi, Gianluca. Streaming active learning strategies for real-life credit card fraud detection: assessment and visualization, International Journal of Data Science and Analytics, 5,4,285-300,2018,Springer International Publishing"
]
}
],
"metadata": {
"authors": [
{
"name": "sekrupa"
}
],
"category": "tutorial",
"compute": [
"AML Compute"
],
"datasets": [
"Creditcard"
],
"deployment": [
"None"
],
"exclude_from_index": false,
"file_extension": ".py",
"framework": [
"None"
],
"friendly_name": "Classification of credit card fraudulent transactions using Automated ML",
"index_order": 5,
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.7"
},
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"tags": [
"AutomatedML"
],
"task": "Classification",
"version": "3.6.7"
},
"nbformat": 4,
"nbformat_minor": 2
}

4
how-to-use-azureml/automated-machine-learning/experimental/classification-credit-card-fraud-local-managed/auto-ml-classification-credit-card-fraud-local-managed.yml Normal file
View File

@@ -0,0 +1,4 @@
name: auto-ml-classification-credit-card-fraud-local-managed
dependencies:
- pip:
- azureml-sdk

435
how-to-use-azureml/automated-machine-learning/experimental/regression-model-proxy/auto-ml-regression-model-proxy.ipynb Normal file
View File

@@ -0,0 +1,435 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copyright (c) Microsoft Corporation. All rights reserved.\n",
"\n",
"Licensed under the MIT License."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"![Impressions](https://PixelServer20190423114238.azurewebsites.net/api/impressions/MachineLearningNotebooks/how-to-use-azureml/automated-machine-learning/experimental/regression-model-proxy/auto-ml-regression-model-proxy.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Automated Machine Learning\n",
"_**Regression with Aml Compute**_\n",
"\n",
"## Contents\n",
"1. [Introduction](#Introduction)\n",
"1. [Setup](#Setup)\n",
"1. [Data](#Data)\n",
"1. [Train](#Train)\n",
"1. [Results](#Results)\n",
"1. [Test](#Test)\n",
"\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Introduction\n",
"In this example we use an experimental feature, Model Proxy, to do a predict on the best generated model without downloading the model locally. The prediction will happen on same compute and environment that was used to train the model. This feature is currently in the experimental state, which means that the API is prone to changing, please make sure to run on the latest version of this notebook if you face any issues.\n",
"This notebook will also leverage MLFlow for saving models, allowing for more portability of the resulting models. See https://docs.microsoft.com/en-us/azure/machine-learning/how-to-use-mlflow for more details around MLFlow is AzureML.\n",
"\n",
"If you are using an Azure Machine Learning Compute Instance, you are all set. Otherwise, go through the [configuration](../../../../configuration.ipynb) notebook first if you haven't already to establish your connection to the AzureML Workspace. \n",
"\n",
"In this notebook you will learn how to:\n",
"1. Create an `Experiment` in an existing `Workspace`.\n",
"2. Configure AutoML using `AutoMLConfig`.\n",
"3. Train the model using remote compute.\n",
"4. Explore the results.\n",
"5. Test the best fitted model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Setup\n",
"\n",
"As part of the setup you have already created an Azure ML `Workspace` object. For Automated ML you will need to create an `Experiment` object, which is a named object in a `Workspace` used to run experiments."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import logging\n",
"\n",
"import json\n",
"\n",
"\n",
"import azureml.core\n",
"from azureml.core.experiment import Experiment\n",
"from azureml.core.workspace import Workspace\n",
"from azureml.core.dataset import Dataset\n",
"from azureml.train.automl import AutoMLConfig"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This sample notebook may use features that are not available in previous versions of the Azure ML SDK."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"print(\"This notebook was created using version 1.40.0 of the Azure ML SDK\")\n",
"print(\"You are currently using version\", azureml.core.VERSION, \"of the Azure ML SDK\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"ws = Workspace.from_config()\n",
"\n",
"# Choose a name for the experiment.\n",
"experiment_name = 'automl-regression-model-proxy'\n",
"\n",
"experiment = Experiment(ws, experiment_name)\n",
"\n",
"output = {}\n",
"output['Subscription ID'] = ws.subscription_id\n",
"output['Workspace'] = ws.name\n",
"output['Resource Group'] = ws.resource_group\n",
"output['Location'] = ws.location\n",
"output['Run History Name'] = experiment_name\n",
"output"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Using AmlCompute\n",
"You will need to create a [compute target](https://docs.microsoft.com/azure/machine-learning/service/concept-azure-machine-learning-architecture#compute-target) for your AutoML run. In this tutorial, you use `AmlCompute` as your training compute resource."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.core.compute import ComputeTarget, AmlCompute\n",
"from azureml.core.compute_target import ComputeTargetException\n",
"\n",
"# Choose a name for your CPU cluster\n",
"# Try to ensure that the cluster name is unique across the notebooks\n",
"cpu_cluster_name = \"reg-model-proxy\"\n",
"\n",
"# Verify that cluster does not exist already\n",
"try:\n",
" compute_target = ComputeTarget(workspace=ws, name=cpu_cluster_name)\n",
" print('Found existing cluster, use it.')\n",
"except ComputeTargetException:\n",
" compute_config = AmlCompute.provisioning_configuration(vm_size='STANDARD_DS12_V2',\n",
" max_nodes=4)\n",
" compute_target = ComputeTarget.create(ws, cpu_cluster_name, compute_config)\n",
"\n",
"compute_target.wait_for_completion(show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Data\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Load Data\n",
"Load the hardware dataset from a csv file containing both training features and labels. The features are inputs to the model, while the training labels represent the expected output of the model. Next, we'll split the data using random_split and extract the training data for the model. "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"data = \"https://automlsamplenotebookdata.blob.core.windows.net/automl-sample-notebook-data/machineData.csv\"\n",
"dataset = Dataset.Tabular.from_delimited_files(data)\n",
"\n",
"# Split the dataset into train and test datasets\n",
"train_data, test_data = dataset.random_split(percentage=0.8, seed=223)\n",
"\n",
"label = \"ERP\"\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Train\n",
"\n",
"Instantiate an `AutoMLConfig` object to specify the settings and data used to run the experiment.\n",
"\n",
"|Property|Description|\n",
"|-|-|\n",
"|**task**|classification, regression or forecasting|\n",
"|**primary_metric**|This is the metric that you want to optimize. Regression supports the following primary metrics: <br><i>spearman_correlation</i><br><i>normalized_root_mean_squared_error</i><br><i>r2_score</i><br><i>normalized_mean_absolute_error</i>|\n",
"|**n_cross_validations**|Number of cross validation splits.|\n",
"|**training_data**|(sparse) array-like, shape = [n_samples, n_features]|\n",
"|**label_column_name**|(sparse) array-like, shape = [n_samples, ], targets values.|\n",
"\n",
"**_You can find more information about primary metrics_** [here](https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-configure-auto-train#primary-metric)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"tags": [
"automlconfig-remarks-sample"
]
},
"outputs": [],
"source": [
"automl_settings = {\n",
" \"n_cross_validations\": 3,\n",
" \"primary_metric\": 'r2_score',\n",
" \"enable_early_stopping\": True, \n",
" \"experiment_timeout_hours\": 0.3, #for real scenarios we recommend a timeout of at least one hour \n",
" \"max_concurrent_iterations\": 4,\n",
" \"max_cores_per_iteration\": -1,\n",
" \"verbosity\": logging.INFO,\n",
" \"save_mlflow\": True,\n",
"}\n",
"\n",
"automl_config = AutoMLConfig(task = 'regression',\n",
" compute_target = compute_target,\n",
" training_data = train_data,\n",
" label_column_name = label,\n",
" **automl_settings\n",
" )"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Call the `submit` method on the experiment object and pass the run configuration. Execution of remote runs is asynchronous. Depending on the data and the number of iterations this can run for a while. Validation errors and current status will be shown when setting `show_output=True` and the execution will be synchronous."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"remote_run = experiment.submit(automl_config, show_output = False)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# If you need to retrieve a run that already started, use the following code\n",
"#from azureml.train.automl.run import AutoMLRun\n",
"#remote_run = AutoMLRun(experiment = experiment, run_id = '<replace with your run id>')"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"remote_run"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Results"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"remote_run.wait_for_completion(show_output=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Retrieve the Best Child Run\n",
"\n",
"Below we select the best pipeline from our iterations. The `get_best_child` method returns the best run. Overloads on `get_best_child` allow you to retrieve the best run for *any* logged metric."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"best_run = remote_run.get_best_child()\n",
"print(best_run)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Show hyperparameters\n",
"Show the model pipeline used for the best run with its hyperparameters."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"run_properties = json.loads(best_run.get_details()['properties']['pipeline_script'])\n",
"print(json.dumps(run_properties, indent = 1)) "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Best Child Run Based on Any Other Metric\n",
"Show the run and the model that has the smallest `root_mean_squared_error` value (which turned out to be the same as the one with largest `spearman_correlation` value):"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"lookup_metric = \"root_mean_squared_error\"\n",
"best_run = remote_run.get_best_child(metric = lookup_metric)\n",
"print(best_run)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"y_test = test_data.keep_columns('ERP')\n",
"test_data = test_data.drop_columns('ERP')\n",
"\n",
"\n",
"y_train = train_data.keep_columns('ERP')\n",
"train_data = train_data.drop_columns('ERP')\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Creating ModelProxy for submitting prediction runs to the training environment.\n",
"We will create a ModelProxy for the best child run, which will allow us to submit a run that does the prediction in the training environment. Unlike the local client, which can have different versions of some libraries, the training environment will have all the compatible libraries for the model already."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.train.automl.model_proxy import ModelProxy\n",
"best_model_proxy = ModelProxy(best_run)\n",
"y_pred_train = best_model_proxy.predict(train_data)\n",
"y_pred_test = best_model_proxy.predict(test_data)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Exploring results"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"y_pred_train = y_pred_train.to_pandas_dataframe().values.flatten()\n",
"y_train = y_train.to_pandas_dataframe().values.flatten()\n",
"y_residual_train = y_train - y_pred_train\n",
"\n",
"y_pred_test = y_pred_test.to_pandas_dataframe().values.flatten()\n",
"y_test = y_test.to_pandas_dataframe().values.flatten()\n",
"y_residual_test = y_test - y_pred_test\n",
"print(y_residual_train)\n",
"print(y_residual_test)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"authors": [
{
"name": "sekrupa"
}
],
"categories": [
"how-to-use-azureml",
"automated-machine-learning"
],
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.2"
}
},
"nbformat": 4,
"nbformat_minor": 2
}

4
how-to-use-azureml/automated-machine-learning/experimental/regression-model-proxy/auto-ml-regression-model-proxy.yml Normal file
View File

@@ -0,0 +1,4 @@
name: auto-ml-regression-model-proxy
dependencies:
- pip:
- azureml-sdk

BIN
how-to-use-azureml/automated-machine-learning/forecasting-backtest-many-models/Backtesting.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 22 KiB

167
how-to-use-azureml/automated-machine-learning/forecasting-backtest-many-models/assets/score.py Normal file
View File

@@ -0,0 +1,167 @@
from typing import Any, Dict, Optional, List
import argparse
import json
import os
import re
import pandas as pd
from matplotlib import pyplot as plt
from matplotlib.backends.backend_pdf import PdfPages
from azureml.automl.core.shared import constants
from azureml.automl.core.shared.types import GrainType
from azureml.automl.runtime.shared.score import scoring
GRAIN = "time_series_id"
BACKTEST_ITER = "backtest_iteration"
ACTUALS = "actual_level"
PREDICTIONS = "predicted_level"
ALL_GRAINS = "all_sets"
FORECASTS_FILE = "forecast.csv"
SCORES_FILE = "scores.csv"
PLOTS_FILE = "plots_fcst_vs_actual.pdf"
RE_INVALID_SYMBOLS = re.compile("[: ]")
def _compute_metrics(df: pd.DataFrame, metrics: List[str]):
"""
Compute metrics for one data frame.
:param df: The data frame which contains actual_level and predicted_level columns.
:return: The data frame with two columns - metric_name and metric.
"""
scores = scoring.score_regression(
y_test=df[ACTUALS], y_pred=df[PREDICTIONS], metrics=metrics
)
metrics_df = pd.DataFrame(list(scores.items()), columns=["metric_name", "metric"])
metrics_df.sort_values(["metric_name"], inplace=True)
metrics_df.reset_index(drop=True, inplace=True)
return metrics_df
def _format_grain_name(grain: GrainType) -> str:
"""
Convert grain name to string.
:param grain: the grain name.
:return: the string representation of the given grain.
"""
if not isinstance(grain, tuple) and not isinstance(grain, list):
return str(grain)
grain = list(map(str, grain))
return "|".join(grain)
def compute_all_metrics(
fcst_df: pd.DataFrame,
ts_id_colnames: List[str],
metric_names: Optional[List[set]] = None,
):
"""
Calculate metrics per grain.
:param fcst_df: forecast data frame. Must contain 2 columns: 'actual_level' and 'predicted_level'
:param metric_names: (optional) the list of metric names to return
:param ts_id_colnames: (optional) list of grain column names
:return: dictionary of summary table for all tests and final decision on stationary vs nonstaionary
"""
if not metric_names:
metric_names = list(constants.Metric.SCALAR_REGRESSION_SET)
if ts_id_colnames is None:
ts_id_colnames = []
metrics_list = []
if ts_id_colnames:
for grain, df in fcst_df.groupby(ts_id_colnames):
one_grain_metrics_df = _compute_metrics(df, metric_names)
one_grain_metrics_df[GRAIN] = _format_grain_name(grain)
metrics_list.append(one_grain_metrics_df)
# overall metrics
one_grain_metrics_df = _compute_metrics(fcst_df, metric_names)
one_grain_metrics_df[GRAIN] = ALL_GRAINS
metrics_list.append(one_grain_metrics_df)
# collect into a data frame
return pd.concat(metrics_list)
def _draw_one_plot(
df: pd.DataFrame,
time_column_name: str,
grain_column_names: List[str],
pdf: PdfPages,
) -> None:
"""
Draw the single plot.
:param df: The data frame with the data to build plot.
:param time_column_name: The name of a time column.
:param grain_column_names: The name of grain columns.
:param pdf: The pdf backend used to render the plot.
"""
fig, _ = plt.subplots(figsize=(20, 10))
df = df.set_index(time_column_name)
plt.plot(df[[ACTUALS, PREDICTIONS]])
plt.xticks(rotation=45)
iteration = df[BACKTEST_ITER].iloc[0]
if grain_column_names:
grain_name = [df[grain].iloc[0] for grain in grain_column_names]
plt.title(f"Time series ID: {_format_grain_name(grain_name)} {iteration}")
plt.legend(["actual", "forecast"])
plt.close(fig)
pdf.savefig(fig)
def calculate_scores_and_build_plots(
input_dir: str, output_dir: str, automl_settings: Dict[str, Any]
):
os.makedirs(output_dir, exist_ok=True)
grains = automl_settings.get(constants.TimeSeries.GRAIN_COLUMN_NAMES)
time_column_name = automl_settings.get(constants.TimeSeries.TIME_COLUMN_NAME)
if grains is None:
grains = []
if isinstance(grains, str):
grains = [grains]
while BACKTEST_ITER in grains:
grains.remove(BACKTEST_ITER)
dfs = []
for fle in os.listdir(input_dir):
file_path = os.path.join(input_dir, fle)
if os.path.isfile(file_path) and file_path.endswith(".csv"):
df_iter = pd.read_csv(file_path, parse_dates=[time_column_name])
for _, iteration in df_iter.groupby(BACKTEST_ITER):
dfs.append(iteration)
forecast_df = pd.concat(dfs, sort=False, ignore_index=True)
# To make sure plots are in order, sort the predictions by grain and iteration.
ts_index = grains + [BACKTEST_ITER]
forecast_df.sort_values(by=ts_index, inplace=True)
pdf = PdfPages(os.path.join(output_dir, PLOTS_FILE))
for _, one_forecast in forecast_df.groupby(ts_index):
_draw_one_plot(one_forecast, time_column_name, grains, pdf)
pdf.close()
forecast_df.to_csv(os.path.join(output_dir, FORECASTS_FILE), index=False)
metrics = compute_all_metrics(forecast_df, grains + [BACKTEST_ITER])
metrics.to_csv(os.path.join(output_dir, SCORES_FILE), index=False)
if __name__ == "__main__":
args = {"forecasts": "--forecasts", "scores_out": "--output-dir"}
parser = argparse.ArgumentParser("Parsing input arguments.")
for argname, arg in args.items():
parser.add_argument(arg, dest=argname, required=True)
parsed_args, _ = parser.parse_known_args()
input_dir = parsed_args.forecasts
output_dir = parsed_args.scores_out
with open(
os.path.join(
os.path.dirname(os.path.realpath(__file__)), "automl_settings.json"
)
) as json_file:
automl_settings = json.load(json_file)
calculate_scores_and_build_plots(input_dir, output_dir, automl_settings)

725
how-to-use-azureml/automated-machine-learning/forecasting-backtest-many-models/auto-ml-forecasting-backtest-many-models.ipynb Normal file
View File

@@ -0,0 +1,725 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Copyright (c) Microsoft Corporation. All rights reserved.\n",
"\n",
"Licensed under the MIT License."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"![Impressions](https://PixelServer20190423114238.azurewebsites.net/api/impressions/MachineLearningNotebooks/how-to-use-azureml/automated-machine-learning/forecasting-hierarchical-timeseries/auto-ml-forecasting-hierarchical-timeseries.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Many Models with Backtesting - Automated ML\n",
"**_Backtest many models time series forecasts with Automated Machine Learning_**\n",
"\n",
"---"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"For this notebook we are using a synthetic dataset to demonstrate the back testing in many model scenario. This allows us to check historical performance of AutoML on a historical data. To do that we step back on the backtesting period by the data set several times and split the data to train and test sets. Then these data sets are used for training and evaluation of model.<br>\n",
"\n",
"Thus, it is a quick way of evaluating AutoML as if it was in production. Here, we do not test historical performance of a particular model, for this see the [notebook](../forecasting-backtest-single-model/auto-ml-forecasting-backtest-single-model.ipynb). Instead, the best model for every backtest iteration can be different since AutoML chooses the best model for a given training set.\n",
"![Backtesting](Backtesting.png)\n",
"\n",
"**NOTE: There are limits on how many runs we can do in parallel per workspace, and we currently recommend to set the parallelism to maximum of 320 runs per experiment per workspace. If users want to have more parallelism and increase this limit they might encounter Too Many Requests errors (HTTP 429).**"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Prerequisites\n",
"You'll need to create a compute Instance by following the instructions in the [EnvironmentSetup.md](../Setup_Resources/EnvironmentSetup.md)."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 1.0 Set up workspace, datastore, experiment"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"gather": {
"logged": 1613003526897
}
},
"outputs": [],
"source": [
"import os\n",
"\n",
"import azureml.core\n",
"from azureml.core import Workspace, Datastore\n",
"import numpy as np\n",
"import pandas as pd\n",
"\n",
"from pandas.tseries.frequencies import to_offset\n",
"\n",
"# Set up your workspace\n",
"ws = Workspace.from_config()\n",
"ws.get_details()\n",
"\n",
"# Set up your datastores\n",
"dstore = ws.get_default_datastore()\n",
"\n",
"output = {}\n",
"output[\"SDK version\"] = azureml.core.VERSION\n",
"output[\"Subscription ID\"] = ws.subscription_id\n",
"output[\"Workspace\"] = ws.name\n",
"output[\"Resource Group\"] = ws.resource_group\n",
"output[\"Location\"] = ws.location\n",
"output[\"Default datastore name\"] = dstore.name\n",
"pd.set_option(\"display.max_colwidth\", None)\n",
"outputDf = pd.DataFrame(data=output, index=[\"\"])\n",
"outputDf.T"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This notebook is compatible with Azure ML SDK version 1.35.1 or later."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"print(\"You are currently using version\", azureml.core.VERSION, \"of the Azure ML SDK\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Choose an experiment"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"gather": {
"logged": 1613003540729
}
},
"outputs": [],
"source": [
"from azureml.core import Experiment\n",
"\n",
"experiment = Experiment(ws, \"automl-many-models-backtest\")\n",
"\n",
"print(\"Experiment name: \" + experiment.name)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 2.0 Data\n",
"\n",
"#### 2.1 Data generation\n",
"For this notebook we will generate the artificial data set with two [time series IDs](https://docs.microsoft.com/en-us/python/api/azureml-automl-core/azureml.automl.core.forecasting_parameters.forecastingparameters?view=azure-ml-py). Then we will generate backtest folds and will upload it to the default BLOB storage and create a [TabularDataset](https://docs.microsoft.com/en-us/python/api/azureml-core/azureml.data.tabular_dataset.tabulardataset?view=azure-ml-py)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# simulate data: 2 grains - 700\n",
"TIME_COLNAME = \"date\"\n",
"TARGET_COLNAME = \"value\"\n",
"TIME_SERIES_ID_COLNAME = \"ts_id\"\n",
"\n",
"sample_size = 700\n",
"# Set the random seed for reproducibility of results.\n",
"np.random.seed(20)\n",
"X1 = pd.DataFrame(\n",
" {\n",
" TIME_COLNAME: pd.date_range(start=\"2018-01-01\", periods=sample_size),\n",
" TARGET_COLNAME: np.random.normal(loc=100, scale=20, size=sample_size),\n",
" TIME_SERIES_ID_COLNAME: \"ts_A\",\n",
" }\n",
")\n",
"X2 = pd.DataFrame(\n",
" {\n",
" TIME_COLNAME: pd.date_range(start=\"2018-01-01\", periods=sample_size),\n",
" TARGET_COLNAME: np.random.normal(loc=100, scale=20, size=sample_size),\n",
" TIME_SERIES_ID_COLNAME: \"ts_B\",\n",
" }\n",
")\n",
"\n",
"X = pd.concat([X1, X2], ignore_index=True, sort=False)\n",
"print(\"Simulated dataset contains {} rows \\n\".format(X.shape[0]))\n",
"X.head()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now we will generate 8 backtesting folds with backtesting period of 7 days and with the same forecasting horizon. We will add the column \"backtest_iteration\", which will identify the backtesting period by the last training date."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"offset_type = \"7D\"\n",
"NUMBER_OF_BACKTESTS = 8 # number of train/test sets to generate\n",
"\n",
"dfs_train = []\n",
"dfs_test = []\n",
"for ts_id, df_one in X.groupby(TIME_SERIES_ID_COLNAME):\n",
"\n",
" data_end = df_one[TIME_COLNAME].max()\n",
"\n",
" for i in range(NUMBER_OF_BACKTESTS):\n",
" train_cutoff_date = data_end - to_offset(offset_type)\n",
" df_one = df_one.copy()\n",
" df_one[\"backtest_iteration\"] = \"iteration_\" + str(train_cutoff_date)\n",
" train = df_one[df_one[TIME_COLNAME] <= train_cutoff_date]\n",
" test = df_one[\n",
" (df_one[TIME_COLNAME] > train_cutoff_date)\n",
" & (df_one[TIME_COLNAME] <= data_end)\n",
" ]\n",
" data_end = train[TIME_COLNAME].max()\n",
" dfs_train.append(train)\n",
" dfs_test.append(test)\n",
"\n",
"X_train = pd.concat(dfs_train, sort=False, ignore_index=True)\n",
"X_test = pd.concat(dfs_test, sort=False, ignore_index=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### 2.2 Create the Tabular Data Set.\n",
"\n",
"A Datastore is a place where data can be stored that is then made accessible to a compute either by means of mounting or copying the data to the compute target.\n",
"\n",
"Please refer to [Datastore](https://docs.microsoft.com/en-us/python/api/azureml-core/azureml.core.datastore(class)?view=azure-ml-py) documentation on how to access data from Datastore.\n",
"\n",
"In this next step, we will upload the data and create a TabularDataset."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.data.dataset_factory import TabularDatasetFactory\n",
"\n",
"ds = ws.get_default_datastore()\n",
"# Upload saved data to the default data store.\n",
"train_data = TabularDatasetFactory.register_pandas_dataframe(\n",
" X_train, target=(ds, \"data_mm\"), name=\"data_train\"\n",
")\n",
"test_data = TabularDatasetFactory.register_pandas_dataframe(\n",
" X_test, target=(ds, \"data_mm\"), name=\"data_test\"\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 3.0 Build the training pipeline\n",
"Now that the dataset, WorkSpace, and datastore are set up, we can put together a pipeline for training.\n",
"\n",
"> Note that if you have an AzureML Data Scientist role, you will not have permission to create compute resources. Talk to your workspace or IT admin to create the compute targets described in this section, if they do not already exist."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Choose a compute target\n",
"\n",
"You will need to create a [compute target](https://docs.microsoft.com/en-us/azure/machine-learning/how-to-set-up-training-targets#amlcompute) for your AutoML run. In this tutorial, you create AmlCompute as your training compute resource.\n",
"\n",
"\\*\\*Creation of AmlCompute takes approximately 5 minutes.**\n",
"\n",
"If the AmlCompute with that name is already in your workspace this code will skip the creation process. As with other Azure services, there are limits on certain resources (e.g. AmlCompute) associated with the Azure Machine Learning service. Please read this [article](https://docs.microsoft.com/en-us/azure/machine-learning/how-to-manage-quotas) on the default limits and how to request more quota."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"gather": {
"logged": 1613007037308
}
},
"outputs": [],
"source": [
"from azureml.core.compute import ComputeTarget, AmlCompute\n",
"\n",
"# Name your cluster\n",
"compute_name = \"backtest-mm\"\n",
"\n",
"\n",
"if compute_name in ws.compute_targets:\n",
" compute_target = ws.compute_targets[compute_name]\n",
" if compute_target and type(compute_target) is AmlCompute:\n",
" print(\"Found compute target: \" + compute_name)\n",
"else:\n",
" print(\"Creating a new compute target...\")\n",
" provisioning_config = AmlCompute.provisioning_configuration(\n",
" vm_size=\"STANDARD_DS12_V2\", max_nodes=6\n",
" )\n",
" # Create the compute target\n",
" compute_target = ComputeTarget.create(ws, compute_name, provisioning_config)\n",
"\n",
" # Can poll for a minimum number of nodes and for a specific timeout.\n",
" # If no min node count is provided it will use the scale settings for the cluster\n",
" compute_target.wait_for_completion(\n",
" show_output=True, min_node_count=None, timeout_in_minutes=20\n",
" )\n",
"\n",
" # For a more detailed view of current cluster status, use the 'status' property\n",
" print(compute_target.status.serialize())"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Set up training parameters\n",
"\n",
"This dictionary defines the AutoML and many models settings. For this forecasting task we need to define several settings including the name of the time column, the maximum forecast horizon, and the partition column name definition. Please note, that in this case we are setting grain_column_names to be the time series ID column plus iteration, because we want to train a separate model for each time series and iteration.\n",
"\n",
"| Property | Description|\n",
"| :--------------- | :------------------- |\n",
"| **task** | forecasting |\n",
"| **primary_metric** | This is the metric that you want to optimize.<br> Forecasting supports the following primary metrics <br><i>normalized_root_mean_squared_error</i><br><i>normalized_mean_absolute_error</i> |\n",
"| **iteration_timeout_minutes** | Maximum amount of time in minutes that the model can train. This is optional but provides customers with greater control on exit criteria. |\n",
"| **iterations** | Number of models to train. This is optional but provides customers with greater control on exit criteria. |\n",
"| **experiment_timeout_hours** | Maximum amount of time in hours that the experiment can take before it terminates. This is optional but provides customers with greater control on exit criteria. |\n",
"| **label_column_name** | The name of the label column. |\n",
"| **max_horizon** | The forecast horizon is how many periods forward you would like to forecast. This integer horizon is in units of the timeseries frequency (e.g. daily, weekly). Periods are inferred from your data. |\n",
"| **n_cross_validations** | Number of cross validation splits. Rolling Origin Validation is used to split time-series in a temporally consistent way. |\n",
"| **time_column_name** | The name of your time column. |\n",
"| **grain_column_names** | The column names used to uniquely identify timeseries in data that has multiple rows with the same timestamp. |\n",
"| **track_child_runs** | Flag to disable tracking of child runs. Only best run is tracked if the flag is set to False (this includes the model and metrics of the run). |\n",
"| **partition_column_names** | The names of columns used to group your models. For timeseries, the groups must not split up individual time-series. That is, each group must contain one or more whole time-series. |"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"gather": {
"logged": 1613007061544
}
},
"outputs": [],
"source": [
"from azureml.train.automl.runtime._many_models.many_models_parameters import (\n",
" ManyModelsTrainParameters,\n",
")\n",
"\n",
"partition_column_names = [TIME_SERIES_ID_COLNAME, \"backtest_iteration\"]\n",
"automl_settings = {\n",
" \"task\": \"forecasting\",\n",
" \"primary_metric\": \"normalized_root_mean_squared_error\",\n",
" \"iteration_timeout_minutes\": 10, # This needs to be changed based on the dataset. We ask customer to explore how long training is taking before settings this value\n",
" \"iterations\": 15,\n",
" \"experiment_timeout_hours\": 0.25, # This also needs to be changed based on the dataset. For larger data set this number needs to be bigger.\n",
" \"label_column_name\": TARGET_COLNAME,\n",
" \"n_cross_validations\": 3,\n",
" \"time_column_name\": TIME_COLNAME,\n",
" \"max_horizon\": 6,\n",
" \"grain_column_names\": partition_column_names,\n",
" \"track_child_runs\": False,\n",
"}\n",
"\n",
"mm_paramters = ManyModelsTrainParameters(\n",
" automl_settings=automl_settings, partition_column_names=partition_column_names\n",
")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Set up many models pipeline"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Parallel run step is leveraged to train multiple models at once. To configure the ParallelRunConfig you will need to determine the appropriate number of workers and nodes for your use case. The process_count_per_node is based off the number of cores of the compute VM. The node_count will determine the number of master nodes to use, increasing the node count will speed up the training process.\n",
"\n",
"| Property | Description|\n",
"| :--------------- | :------------------- |\n",
"| **experiment** | The experiment used for training. |\n",
"| **train_data** | The file dataset to be used as input to the training run. |\n",
"| **node_count** | The number of compute nodes to be used for running the user script. We recommend to start with 3 and increase the node_count if the training time is taking too long. |\n",
"| **process_count_per_node** | Process count per node, we recommend 2:1 ratio for number of cores: number of processes per node. eg. If node has 16 cores then configure 8 or less process count per node or optimal performance. |\n",
"| **train_pipeline_parameters** | The set of configuration parameters defined in the previous section. |\n",
"\n",
"Calling this method will create a new aggregated dataset which is generated dynamically on pipeline execution."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.contrib.automl.pipeline.steps import AutoMLPipelineBuilder\n",
"\n",
"\n",
"training_pipeline_steps = AutoMLPipelineBuilder.get_many_models_train_steps(\n",
" experiment=experiment,\n",
" train_data=train_data,\n",
" compute_target=compute_target,\n",
" node_count=2,\n",
" process_count_per_node=2,\n",
" run_invocation_timeout=920,\n",
" train_pipeline_parameters=mm_paramters,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.pipeline.core import Pipeline\n",
"\n",
"training_pipeline = Pipeline(ws, steps=training_pipeline_steps)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Submit the pipeline to run\n",
"Next we submit our pipeline to run. The whole training pipeline takes about 20 minutes using a STANDARD_DS12_V2 VM with our current ParallelRunConfig setting."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"training_run = experiment.submit(training_pipeline)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"training_run.wait_for_completion(show_output=False)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Check the run status, if training_run is in completed state, continue to next section. Otherwise, check the portal for failures."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 4.0 Backtesting\n",
"Now that we selected the best AutoML model for each backtest fold, we will use these models to generate the forecasts and compare with the actuals."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Set up output dataset for inference data\n",
"Output of inference can be represented as [OutputFileDatasetConfig](https://docs.microsoft.com/en-us/python/api/azureml-core/azureml.data.output_dataset_config.outputdatasetconfig?view=azure-ml-py) object and OutputFileDatasetConfig can be registered as a dataset. "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.data import OutputFileDatasetConfig\n",
"\n",
"output_inference_data_ds = OutputFileDatasetConfig(\n",
" name=\"many_models_inference_output\",\n",
" destination=(dstore, \"backtesting/inference_data/\"),\n",
").register_on_complete(name=\"backtesting_data_ds\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"For many models we need to provide the ManyModelsInferenceParameters object.\n",
"\n",
"#### ManyModelsInferenceParameters arguments\n",
"| Property | Description|\n",
"| :--------------- | :------------------- |\n",
"| **partition_column_names** | List of column names that identifies groups. |\n",
"| **target_column_name** | \\[Optional\\] Column name only if the inference dataset has the target. |\n",
"| **time_column_name** | Column name only if it is timeseries. |\n",
"| **many_models_run_id** | \\[Optional\\] Many models pipeline run id where models were trained. |\n",
"\n",
"#### get_many_models_batch_inference_steps arguments\n",
"| Property | Description|\n",
"| :--------------- | :------------------- |\n",
"| **experiment** | The experiment used for inference run. |\n",
"| **inference_data** | The data to use for inferencing. It should be the same schema as used for training.\n",
"| **compute_target** | The compute target that runs the inference pipeline.|\n",
"| **node_count** | The number of compute nodes to be used for running the user script. We recommend to start with the number of cores per node (varies by compute sku). |\n",
"| **process_count_per_node** | The number of processes per node.\n",
"| **train_run_id** | \\[Optional\\] The run id of the hierarchy training, by default it is the latest successful training many model run in the experiment. |\n",
"| **train_experiment_name** | \\[Optional\\] The train experiment that contains the train pipeline. This one is only needed when the train pipeline is not in the same experiement as the inference pipeline. |\n",
"| **process_count_per_node** | \\[Optional\\] The number of processes per node, by default it's 4. |"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.contrib.automl.pipeline.steps import AutoMLPipelineBuilder\n",
"from azureml.train.automl.runtime._many_models.many_models_parameters import (\n",
" ManyModelsInferenceParameters,\n",
")\n",
"\n",
"mm_parameters = ManyModelsInferenceParameters(\n",
" partition_column_names=partition_column_names,\n",
" time_column_name=TIME_COLNAME,\n",
" target_column_name=TARGET_COLNAME,\n",
")\n",
"\n",
"inference_steps = AutoMLPipelineBuilder.get_many_models_batch_inference_steps(\n",
" experiment=experiment,\n",
" inference_data=test_data,\n",
" node_count=2,\n",
" process_count_per_node=2,\n",
" compute_target=compute_target,\n",
" run_invocation_timeout=300,\n",
" output_datastore=output_inference_data_ds,\n",
" train_run_id=training_run.id,\n",
" train_experiment_name=training_run.experiment.name,\n",
" inference_pipeline_parameters=mm_parameters,\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.pipeline.core import Pipeline\n",
"\n",
"inference_pipeline = Pipeline(ws, steps=inference_steps)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"inference_run = experiment.submit(inference_pipeline)\n",
"inference_run.wait_for_completion(show_output=False)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 5.0 Retrieve results and calculate metrics\n",
"\n",
"The pipeline returns one file with the predictions for each times series ID and outputs the result to the forecasting_output Blob container. The details of the blob container is listed in 'forecasting_output.txt' under Outputs+logs. \n",
"\n",
"The next code snippet does the following:\n",
"1. Downloads the contents of the output folder that is passed in the parallel run step \n",
"2. Reads the parallel_run_step.txt file that has the predictions as pandas dataframe \n",
"3. Saves the table in csv format and \n",
"4. Displays the top 10 rows of the predictions"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from azureml.contrib.automl.pipeline.steps.utilities import get_output_from_mm_pipeline\n",
"\n",
"forecasting_results_name = \"forecasting_results\"\n",
"forecasting_output_name = \"many_models_inference_output\"\n",
"forecast_file = get_output_from_mm_pipeline(\n",
" inference_run, forecasting_results_name, forecasting_output_name\n",
")\n",
"df = pd.read_csv(forecast_file, delimiter=\" \", header=None, parse_dates=[0])\n",
"df.columns = list(X_train.columns) + [\"predicted_level\"]\n",
"print(\n",
" \"Prediction has \", df.shape[0], \" rows. Here the first 10 rows are being displayed.\"\n",
")\n",
"# Save the scv file with header to read it in the next step.\n",
"df.rename(columns={TARGET_COLNAME: \"actual_level\"}, inplace=True)\n",
"df.to_csv(os.path.join(forecasting_results_name, \"forecast.csv\"), index=False)\n",
"df.head(10)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## View metrics\n",
"We will read in the obtained results and run the helper script, which will generate metrics and create the plots of predicted versus actual values."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from assets.score import calculate_scores_and_build_plots\n",
"\n",
"backtesting_results = \"backtesting_mm_results\"\n",
"os.makedirs(backtesting_results, exist_ok=True)\n",
"calculate_scores_and_build_plots(\n",
" forecasting_results_name, backtesting_results, automl_settings\n",
")\n",
"pd.DataFrame({\"File\": os.listdir(backtesting_results)})"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The directory contains a set of files with results:\n",
"- forecast.csv contains forecasts for all backtest iterations. The backtest_iteration column contains iteration identifier with the last training date as a suffix\n",
"- scores.csv contains all metrics. If data set contains several time series, the metrics are given for all combinations of time series id and iterations, as well as scores for all iterations and time series ids, which are marked as \"all_sets\"\n",
"- plots_fcst_vs_actual.pdf contains the predictions vs forecast plots for each iteration and, eash time series is saved as separate plot.\n",
"\n",
"For demonstration purposes we will display the table of metrics for one of the time series with ID \"ts0\". We will create the utility function, which will build the table with metrics."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"def get_metrics_for_ts(all_metrics, ts):\n",
" \"\"\"\n",
" Get the metrics for the time series with ID ts and return it as pandas data frame.\n",
"\n",
" :param all_metrics: The table with all the metrics.\n",
" :param ts: The ID of a time series of interest.\n",
" :return: The pandas DataFrame with metrics for one time series.\n",
" \"\"\"\n",
" results_df = None\n",
" for ts_id, one_series in all_metrics.groupby(\"time_series_id\"):\n",
" if not ts_id.startswith(ts):\n",
" continue\n",
" iteration = ts_id.split(\"|\")[-1]\n",
" df = one_series[[\"metric_name\", \"metric\"]]\n",
" df.rename({\"metric\": iteration}, axis=1, inplace=True)\n",
" df.set_index(\"metric_name\", inplace=True)\n",
" if results_df is None:\n",
" results_df = df\n",
" else:\n",
" results_df = results_df.merge(\n",
" df, how=\"inner\", left_index=True, right_index=True\n",
" )\n",
" results_df.sort_index(axis=1, inplace=True)\n",
" return results_df\n",
"\n",
"\n",
"metrics_df = pd.read_csv(os.path.join(backtesting_results, \"scores.csv\"))\n",
"ts = \"ts_A\"\n",
"get_metrics_for_ts(metrics_df, ts)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Forecast vs actuals plots."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from IPython.display import IFrame\n",
"\n",
"IFrame(\"./backtesting_mm_results/plots_fcst_vs_actual.pdf\", width=800, height=300)"
]
}
],
"metadata": {
"authors": [
{
"name": "jialiu"
}
],
"categories": [
"how-to-use-azureml",
"automated-machine-learning"
],
"kernelspec": {
"display_name": "Python 3.6",
"language": "python",
"name": "python36"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.9"
}
},
"nbformat": 4,
"nbformat_minor": 4
}

4
how-to-use-azureml/automated-machine-learning/forecasting-backtest-many-models/auto-ml-forecasting-backtest-many-models.yml Normal file
View File

@@ -0,0 +1,4 @@
name: auto-ml-forecasting-backtest-many-models
dependencies:
- pip:
- azureml-sdk

Some files were not shown because too many files have changed in this diff Show More